# sensho Personal site of sensho — engineering at a startup, with a background in competitive esports (Valorant, Fortnite), content creation (finance TikTok), Roblox game development, and Software Engineering at the University of Ottawa (currently on leave). Contact: - Email: sensho@sensho.xyz - Twitter/X: https://twitter.com/sensho (@sensho) - Discord: xa8 - MyAnimeList: https://myanimelist.net/animelist/brobc --- ## Home (https://sensho.xyz/) Headline: "hi, i'm Om" ### Me I am curious ### Present - **latitude.io**: I do product, applied AI (agents, games, etc.), platform, monetization, and other stuff. Backed by Google's AI Futures Fund, Roblox's co-founder, Midjourney, and more - **autonomous research** — https://sensho.xyz/autonomous-research: Applied AI to solve frontier research problems - **gud labs** (soon): Simulating world environments for agents --- ## /autonomous-research — https://sensho.xyz/autonomous-research *4/21/2026* > applied ai to solve frontier research problems ### Papers - **Improved Bounds for the Primitive-Set Saturation Game (Erdős Problem 872)** (April 21, 2026) — https://sensho.xyz/papers/erdos-872.pdf ### Work - [erdos-co-researcher](https://github.com/xa8zz/erdos-co-researcher) — A basic co-researcher agent to help automate math research ### Full post ## autonomous research ![autonomous research program chart](/blog/program-chart.svg) Fully autonomous research systems are inevitable. The state of AI-driven research today is where AI for coding was 8 to 12 months ago. Inconsistent moments of amazement with spiky task and prompt dependent performance. Right now, AI-driven research is very primitive. You need a human handling the systems end-to-end, similar to what coding is to software engineering. Most AI-driven research today is from humans fitting the AI within a narrow enough problem space for them to make progress. The mistake that some people make here, is assuming that humans will always be necessary for this and there is some kind of abstract "research taste" that will be necessary in the near term future to do SOTA research. We are at a temporary moment in time where humans are needed to get the most out of these systems as they scale and iterate, but said scaling and iterating is converging on replacing the handler - not the other way around. When I do research with AIs, this is what I provide them: - Mistake analysis - Verification - Coordination - Compute access - Process - Research Strategy - Continual learning over long contexts - Pattern recognition on long horizons (basically all of the above haha) I am not providing any value that cannot eventually be solved. Everything that I am doing and more is verifiable and will therefore be solved. Some are compute constrained, some are capability constrained, some are RL resource constrained, some are tooling constrained - but in this industry, a constraint has almost always implied solvability. After my experience using these systems, I do not think an autonomous frontier research agent can exist today. Most labs or companies claiming otherwise right now have not built a true harness. Most are using hack-y harnesses that force certain behaviors that do not scale or generalize with model behavior. In the near future, the best harnesses are those that use minimal task-specific tooling, the ideal harness for coding will also be the ideal harness for research, marketing, etc. The greatest lesson of my harness engineering endeavours is that a generalized harness is what the future will seek. This is because as we race to make every domain in society verifiable, we want the harness used to hit verifiable targets to be generalized as well (including for research). It's very likely that the great task for diffusion, acceleration, society, etc. over the coming decade with AI advances will be making *all* domains verifiable to AI systems to fit this trend line of how real progress gets made with AI. You are doing your harness engineering correctly if, when a model upgrade comes and you swap it in, that the system is immediately upgraded and then extensible by just adding more tools, context, and verifiability on its work rather than changing the core primitives the system operates on. This is similar to the vast swath of coding agents made in 2024 that have all now centralized to a core agent tool loop + file tree system. Therefore, in my personal research ventures I apply the same base primitives that coding agents use and then just expand upon them: AI requires scale by its very nature, even with every algorithmic advancement in the world it is the great polarizer of any distribution. The best AI systems will only be more centralized and resource intensive going forward. Attempting to argue that the best near term systems will prefer different harnesses is a fight against the centralizing and verification tendencies of this entire domain by its nature. File trees are a horribly inefficient way for models to build continual learning - but it works with human guidance (for now) and can generalize a model's ability far more than its in-context learning window and is already existing proof that the value that humans provide to these systems is going to diminish with time as we find more processes to eliminate the work that we do. The design is pretty simple, a continuous orchestrator that follows the repo's guidelines and principles to synthesize and analyze the work so far. There is nothing a part of the research program that does not exist in the file tree. If anything exists outside of the file tree it is lost forever. Until we get unlimited and perfect attention algorithms for context (or continual learning) this must be a core principle for any system. If it exists in the file tree you apply many fun agent techniques over it to make the research process streamlined. It has enabled helpful techniques like front-matter, skills, third opinions, custom tools, etc. that help make the manual process that humans do prompting GPT Pro in ChatGPT easier. These are pretty basic primitives and nothing revolutionary is done here - because there doesn't need to be. The models are not trained to run on your custom research loop that guides it in a certain way to approach research, that will not work long-term and just be deprecated as model capabilities expand. I'd like to write a long piece on my experience using the systems today and the specific weaknesses but I don't think that will be valuable for long. I think the best assessment will be, for people like me who cannot prove any of the math in the papers I write myself, what value do I provide? Those will be the best tell of the current bottlenecks and work to be done - which labs are already racing to (attempt) to do. I can envision a day <1 year where a mid tier unsolved Erdős-level problem is put into a box and a lean verified solution returns later. Math, and to some extent, STEM research, is one of the easier ones because it is arguably even more verifiable than most of the day-to-day consumer uses for AI. This doesn't mean I don't think the other problems won't get solved either; simulated worlds, scaling RL (and letting different environments generalize), etc. there are many roads to solving more "abstractly" verified domains. It sounds cringe to write out and sounded cooler in my head but I can definitely see a world where all "first-time events" in the modern day like an AI in a position of government, running businesses, etc. have all been simulated countless times in worlds against trillions of AI agents recursively improving themselves through RL and the best one wins out and comes to human land. Lol. What a beautiful world we live in where I can write that and it's a genuine possibility. We should realize that the bottleneck to progress is how to conform our human systems to these, not the other way around, and then operate on that principle for future changes. I'm sorry that I have very little to personally offer to frontier research, if I did maybe I could help get my point across better. But I do hope that my work helps at least someone see the value in utilizing these systems to advance humanity in any domain and the good that can come of it. If you think there is a mistake in any of my work or there is some major way to improve please email me at sensho@sensho.xyz, I will try my best to fix it. --- ## /esports — https://sensho.xyz/esports *8/12/2025* > Peaked top 25 NA in valorant, 8x radiant, CVAL record holder, top 100 in Fortnite ### Highlights ### valorant - peaked top 25 NA - 7x radiant; been top 25, 50, 100 to 300 etc. multiple times - played CVAL (Riot Games' official collegiate circuit w/ $500k+ in prizing), made playoffs and deep runs, record for highest ACS in CVAL league - multiple scholarship and t2 pro offers ### fortnite - top 100 FNCS, multiple top 200 to 500 FNCS finishes - 20x+ top 1000 tournament finishes ### Full post ## Valorant I started out as iron in Valorant, the lowest rank in the game. It took me a month to get to platinum, four months to immortal, and a year+ to Radiant. It was my first ever competitive FPS. I mostly played Valorant because of how addictive grinding ranked was to my competitive desires. The ranked system in the game is effective, so over ~3 years it ate up free time. I learned a lot about the process of improving, hit multiple what I thought were "ceilings," and surpassing them. Also, playing against (and farming) pro players and popular streamers in ranked is also very fun :p. I had no plans or desire to ever pursue competitive Valorant beyond ranked as it doesn't align with what I wanna do long-term, it was just a competitive outlet for me. Playing pro and competitive at a high level is typically a 12 hours/day career and I wouldn't be able to fit my life desires if I was doing that for years. This was until my school friends asked me (a lot) to join the college Valorant team and apply my skills there. I joined on the condition that we wouldn't be scrimmaging or having practices that take up lots of time outside of matches; they agreed so I played. I was playing CVAL, which is considered Tier 2 valorant: basically a level below the official professional circuit with large salaries and such. In the last semester of CVAL I played, I set the record for the highest ACS (combat score, measure of your impact) during playoffs. I was elite at it and the matches in of themselves were extremely enjoyable. Both teams are fully trying, lots of practice (from the other team at least) with set strategies, warm-ups, etc. and you lay it all down in a Best of 3 or 5 series that takes hours. Our team made some decent runs and had success, and I had good teammates that set up to enable me well. I also played in a couple of LANs, winning one of them with some fun memories doing so. As of writing this I'm still Radiant top 300, but that's only because this ranked season hasn't ended yet. I've only played 4-5 matches in the past month and don't intend on playing actively again as it draws my attention from other things I want to do. I've had outreach from schools with scholarship opportunities and T2 team trial offers. Questions from friends or Valorant players I know are around why I haven't tried pursuing a pro career if I'm able to play and excel at that level; it's weird to people when I try to explain my lack of interest in doing so. I understand why it can seem weird and don't blame them for thinking that way, it's just how I go about life :P. The natural tendencies that let me have fun and do well in Valorant (without getting burnt out unlike other activities) are the same tendencies that urge me to go elsewhere for long-term ventures, and I'm fine with that. ## Fortnite Fortnite was my first competitive shooter! Prior to this I had mainly played Roblox, Minecraft, and strategy/rts games. People like to clown on Fortnite competitively as it's a tier 2 esport and considered a childish game. I disagree, I actually think Fortnite is one of the hardest competitive shooters out there. The building mechanic adds a huge skill gap that can't be pre-learned/transferred from other games. This means anyone good at Fortnite actually needs to put in effort to become good at building, editing, aiming, and all other mechanics. The tournaments are also open to play by anyone and online, so access is easy and meritocratic. I got into Fortnite during OG Fortnite in 2018; I played with my school friends and had a blast. I continued playing as the OG hype died down and got into playing tournaments/Fortnite esports with a few friends. It was similar to Valorant — just a side hobby during early high school. I had some tournament earnings and decent placements but my success was lower relative to Valorant. In both games though I never cared for competing in esports much, it was just a competitive hobby that I got good at and when you get skilled you just get drawn to esports naturally. In Fortnite my playstyle was known for being very mechanics heavy; I'd mastered building, editing, and aiming through grinding Creative mode and would just win lots of fights in tournament matches. Eventually after hitting a ceiling I was able to learn better game sense and improve that, peaking around Top 100 and 200 finishes in FNCS (main esports event). Despite my esports involvement in these games, I'm probably best/most-passionate towards strategy games. Problem is they don't have a ranked system, but I have lots of hours over lots of different strategy/RTS type games — those kinds of games are my true passions and the games I love beyond simply competing. My main draw to competitive shooters isn't that I love the base game as much but rather the idea of a meritocratic, competitive playground with a well-run ranked system that quantifies your skill and rewards you for it. --- ## /tiktok — https://sensho.xyz/tiktok *8/12/2025* > Was a tiktoker about finance content. 195k followers and 50m+ views in ~2 months ### Highlights - end of '21 to early '22 I was a content creator on tiktok - grew a finance tiktok account from 0 to 195k followers and 50m+ views in <2 months - focused heavily on content strategy: studying videos, retention, formats, storytelling, viral design, etc. basically tried memetic theory applied to tiktok - ran a Discord server for the account with 9k members ### Full post I had some smaller content creation success prior to this on most social medias but this one was the first one that had videos reach the mainstream.. I got into it because I was seeing lots of content in this niche on my feed and thought all of it sucked and could be done better. If I look back on my videos now 3-4 years later, they also suck and are cringe but for the time had some innovative formats and funnels that I had fun operating. The bizdev part of the business was also enjoyable; operating the background monetization and operations was like an optimization game. The part that I initially was looking forward to the most was meeting likeminded people like myself. I thought people in this industry were also curious about video formats, finance theory, etc. Unfortunately, that did not happen much. There were interesting people, but a lot of the people in that content niche are what you'd expect from finance/crypto content creators. More of the grifty and greedy archetype who were not enjoyable people to interact with for someone who does things out of interest/curiosity. I don't even have a problem with the greed, I understand it; it was more the lack of passion that made interactions uninteresting. After my initial optimism I eventually stopped talking to others in that creator space, and just focused on my own content. Eventually, because my passion/interest for this field had been stretched into making baity/grifty content and sometimes needing to interact with people I didn't really care for it became boring to me. On top of this you apply the status value/expectations and everyone around me IRL telling me how great this is despite my dismay. I enjoyed distribution, memetic theory, growth hacking, system/constraint optimization, etc.; I liked everything about the business but not what the business was actually producing (the content and people I met). I've gotten told I should've kept going for financial success, but I was thankfully in a position to not need to. College was starting up near this point and I got to enjoy the remainder of the last summer I had before school. In early 2023 I got back into it for a month for two main reasons: 1) I had a friend that thought my earlier success with this was purely luck and I wanted to disprove him. 2) I wanted to see if maybe I'd enjoy the content itself this time and that things could have changed. I started posting on the account again and succeeded in disproving my friend. The second hope did not go well though, as I still found the industry grifty and cringe. I considered pivoting into a different niche or changing up content but I actually started enjoying the other projects I was working on a lot more at the time. Since this project I haven't attempted content/distribution ventures like this and only post on social media for personal enjoyment, but this experience was extremely formative for me. Beyond just what I got good at while running the operation, I learned a lot about what I like, dislike, and schools of thought around fulfillment. It's helped me learn what I actually like doing now. Some of the stuff that I learned here around growth, distribution, and business development I still enjoy — but only in combination with the other stuff I've learned that I like! :) --- ## /roblox-dev — https://sensho.xyz/roblox-dev *8/12/2025* > Ran a roblox game dev group during COVID that had its games played 12 million+ times ### Highlights - start of COVID to end of 2020 i ran a roblox game studio that built games - generated 12M game visits across multiple games - led roblox groups, devs, community, support, etc. ### Full post I played a large amount of Roblox growing up and did a variety of things on the platform. It taught me a lot and gave me fun memories with friends and learning skills early on that I would've never been able to know otherwise. During the mentioned time, I combined a lot of the skills I'd picked up in the Roblox ecosystem to run a Roblox game studio; it's given me lasting memories because I got to involve friends in the operation. This was a project that I ran during school and was mainly a way for my group of friends to work on something together. I led most of the efforts: directing our development and roadmap. We initially made a few games on our own that did well, hitting the front page of some Roblox game categories (for a short period). The games we made weren't that innovative or good (they were slop in retrospect, lol). We just did remixes and iterations of currently existing game formats; this made development time and updating much easier. This was a deliberate choice that we made knowing what kinds of games people liked on Roblox. After we had some success in our games, the funds were used to get involved in other upcoming games. This was because through me and friends, we knew lots of other people we found cool working on their own games. Some of them were newer devs, that e.g., needed help with scripting or didn't want to manage things like monetization or communities. We would join game projects we found interesting and help with technical or operational aspects. It sounds generic but genuinely my favroite part was making updates/changes and hoping players would react well. It involved parts like developing fast in of itself but also methodologies on finding out what users wanted. This was part of the reasoning behind not trying to make the game unnecessarily complex or innovative without user feedback. The goal was to iterate through their feedback quickly. The most successful game in Roblox history "Grow a Garden" which was released in 2025 also utilizes this concept of simplicity. Towards the end of 2020 I had moved on to some other interests, so I passed leading our group to one of my close friends that was highly competent. After a few more months, our game studio was eventually was disbanded; this wasn't due to the new leader but rather that most of our friend group had grown more experienced, older, and connected, which led to everyone having their own self-interests in Roblox to pursue. I'm not upset over that, rather I'm happy our project was fruitful and let us have a large impact on Roblox. The close friend that I had let lead is now one of the biggest Roblox developers, with many others going to do cool things in the ecosystem as well!. --- ## /education — https://sensho.xyz/education *8/12/2025* > Software engineering at uOttawa (on leave) ### Highlights - attending the University of Ottawa for Software Engineering, on temporary leave - accepted to all major Canadian schools (Waterloo, UofT, others) - scholarships to all major schools - 98% high school grade average for school applications ### Full post I don't have too much to write here unlike the other sections. I am grateful to live a life where my curiosity alone is enough to get me far in anything that I want to do in the most exciting time in human history. I think that school can be valuable and is what you make of it, I prefer to do that in random areas that I get interested in instead. It's not really because I work hard or that I'm vdisciplined (rather, the opposite), but that I'm acting in my own self-interest because I genuinely love what I do and get to think about more than I could at a school.