Software Engineer, Agents
Design and build agentic systems for AI-native video creation, integrating LLMs and evaluation frameworks to power creative workflows. Requires 5+ years building ML/agentic systems in production.
A YC member benefit, not a candidate product
Work at a Startup runs as portfolio infrastructure for Y Combinator. It exists to funnel candidate applications into YC's roughly 1,000 backed companies, with a single profile that lets you submit to hundreds of roles in a click. That mechanic is great for YC, because it concentrates supply into their portfolio. It has become less great for candidates, because the same mechanic is what poisons the queue.
The product has barely evolved. One Hacker News post famously framed it as a rarely mentioned benefit of YC, which tells you how much engineering attention it receives compared to YC's main product. There is no public API. The application flow forces every user to pick a single role bucket via a radio button that locks the rest of the account into that category. Hotfix is the broader-startup alternative built for candidates, not for a portfolio.
Why people leave
The most structural complaint is the scope cap. Every company on Work at a Startup is YC-backed, by design. For a candidate whose taste was formed by YC companies but who would also take a great seed-stage role at a non-YC startup, the entire frontier-lab cluster, the design-led non-YC studios, and the well-run bootstrapped category are all hidden by the funnel. The constraint is a feature for YC and a hard ceiling for you.
The application mechanic is the second complaint. One profile, hundreds of jobs, founders see your application directly. That value proposition has collapsed under its own success. A YC founder on a recent Ask HN thread reported 172 daily applicants, of which 22 looked like real humans and 150 were primarily AI-generated. Candidates describe applying to twenty or thirty roles and hearing back from zero, calling the experience sending applications into oblivion. The volume mechanic is what enables the AI spam, and the AI spam is what makes founders stop reading.
The third complaint is product mechanics. The radio-button role lock means you can be good at frontend or devops, but not both, on a single account. The profile gates value, so casual browsing is friction. There is no real-time alerting for new roles that match your preferences. Listings are slow to refresh because the system optimizes for the founder pipeline, not for the candidate's hunting cycle.
Stacked together, these are not three small frustrations. They describe a product built as portfolio infrastructure for one investor, with all the constraints that come with that, and none of the candidate-side product investment a real job board would have made years ago.
Side by side
| Dimension | Work at a Startup | Hotfix |
|---|---|---|
| Catalog scope | YC-backed companies only | Handpicked startups across the wider tech ecosystem |
| Browse without account | Profile required for full access | Full browse without signing up |
| Application mechanic | One profile, blast to hundreds of YC roles | Apply directly to roles that actually fit |
| Role flexibility | Single-bucket radio button per account | No role lock-in, follow whatever fits |
| Job alerts | No real-time alerts | Real-time, preference-driven |
| Listing freshness | Optimized for founder pipeline, slow refresh | Pulled the day the company closes the role |
| Price for candidates | Free | Free |
The trade we made
Work at a Startup optimizes for funnel efficiency, concentrating every YC company's hiring into one queue. Hotfix optimizes for fit, surfacing the roles where the company and the candidate actually match, regardless of who wrote the seed check. If you only want YC companies, Work at a Startup is the right tool. If your taste also includes the seed-stage spinouts from those companies, the not-YC frontier labs, and the wider category of well-run startups, the YC cap is doing real work against you.
Volume is not the same as fit. The value proposition of Work at a Startup is volume: one profile, hundreds of YC roles, founders see your application directly. That has become its biggest weakness, because every YC company is now competing for attention in the same queue, and the queue is overrun with the AI spam the mechanic enables. Hotfix's bet is that browsing without a login wall and applying to specific roles you care about beats blasting a profile into a single shared funnel.
The page you are on is the comparison. The actual product is below: live tech and startup roles right now, no signup required to browse.
Honest assessment
Pick the right tool
Proof, not pitch
A sample of roles posted recently. Browse the full set without an account.
Design and build agentic systems for AI-native video creation, integrating LLMs and evaluation frameworks to power creative workflows. Requires 5+ years building ML/agentic systems in production.
Own Core Infra team end-to-end: set priorities, raise execution quality, and turn ICs into a self-organizing team. Requires 8+ years building production software (4+ in startups) managing platform engineers, plus deep familiarity with high-throughput and AI-powered systems.
Develop post-training recipes and build enterprise data agents for autonomous planning, code generation, and multi-step workflows. Requires 2+ years applied research experience shipping prototypes, plus expertise in LLMs, agents, and RL.
Senior engineer on the R&D team building production infrastructure for LLM agents, browser agents, and self-improving harnesses. Owns the full lifecycle from research through customer delivery.
Full-stack engineer building Estimate, a product that converts generative building designs into instant material takeoffs. Owns projects end-to-end across frontend, backend, and geometry layers.
Product Designer shaping pilot interactions with AI cockpit systems. Own end-to-end design from research through specs, build design systems, and translate complex AI data into safety-critical interfaces.
Research Scientist II building and improving fraud risk models and scam detection systems using audio, behavioral, and metadata signals. Requires an advanced degree and 3+ years of applied ML experience with Python and modern ML frameworks.
Designs model behavior and human-AI interactions, collaborating with researchers and PMs to improve model responses and user experience. Requires strong taste, writing, and technical intuition.