ML Engineer

Visia; New York (NY); Ibrido; USD 170.000 - 250.000; Tempo pieno

Vantaggi offerti da questo lavoro

Employer premiums

401k match

Equity with 10-year window

Unlimited PTO

Descrizione del lavoro

Visia in New York City is seeking a senior ML engineer to own and enhance the vision system’s learning agent. You will post-train open-weight models (SFT, RL methods) and build tools and evals to improve answers across 50+ industrial sites.

Work in a hybrid setup, three days a week in the Union Square area. You’ll expand the tool catalog, manage long-context sessions, and push models beyond current weight limits with routing and planner systems.

Competenze

• Shipped a tool-calling LLM or VLM to real users and can explain evaluation outcomes.
• Post-trained open-weight LLMs or VLMs with SFT and RL methods, and can explain the impact on eval results.

Mansioni

• Post-train open-weight models on our tasks: SFT on curated traces, then RL with verifiable rewards, and build the reward functions, rollout environment, and eval gate.
• Own the agent harness: context management over long multi-turn sessions, tool-call budgets, background tools, failure recovery, and the policy deciding which tool calls need approval.
• Grow the tool catalog (SQL over Postgres and ClickHouse, frame retrieval, segmentation, depth, model training and evaluation), and give each tool a typed schema, clear errors, and a test.
• Run the evaluation loop: replay real questions against a candidate harness or model, grade the answers with calibrated judges, and report quality and dollars per correct answer before anything ships.
• Push open-source models beyond their weight class with routing, test-time selection, and advisor and planner/worker setups where a strong model plans and cheaper ones execute.

Conoscenze

Tool-calling LLM

Open-weight LLM

RL methods

Strumenti

PostgreSQL

ClickHouse

LLM tooling

Descrizione del lavoro

Visia | New York, NY | Hybrid | Full-time

Base salary: $170,000–$250,000 per year

You'll own the agent that trains, evaluates, and improves the models behind our vision systems at 50+ industrial sites. You'll post-train the open-weight models that power it (SFT, GRPO, and related methods), build its tools and evals, and get the best possible answers at every point on the cost curve.

The role

Our vision systems run at steel mills, scrap yards, and recycling plants. Behind them, our vision agent Atlas trains and evaluates our models, finds their weaknesses, and searches millions of images for examples of rare classes.

It is already our main way to investigate and iterate on our deployed models and our research. It works, but it can become a much more powerful tool.

This is not a prompt-engineering role. Most of the work is systems and measurement: deciding what the agent can see and do, and proving on a replayed set of real questions that a change improved its answers.

Your teammates already contribute comfortably in a mix of programming languages, and we expect you to do the same. You'll work in person about 3 days a week at our office near Union Square, NYC.

What you'll do

Post-train open-weight models on our own tasks: SFT on curated traces, then RL with verifiable rewards (GRPO and its variants, and DPO-style preference methods where they fit). Build the reward functions, the rollout environment, and the eval gate, then serve the result.

Own the agent harness: context management over long multi-turn sessions, tool-call budgets, background tools, failure recovery, and the policy that decides which tool calls need approval.

Grow the tool catalog (SQL over Postgres and ClickHouse, frame retrieval, segmentation, depth, model training and evaluation), and give each tool a typed schema, clear errors, and a test.

Run the evaluation loop: replay real questions against a candidate harness or model, grade the answers with calibrated judges, and report quality and dollars per correct answer before anything ships.

Push open-source models beyond their weight class with routing, test-time selection, and advisor and planner/worker setups in which a strong model plans and cheap ones execute.

Must-haves

• You've shipped a tool-calling LLM or VLM agent to real users, and you can explain how you evaluated it and how it failed.
• You've post-trained open-weight LLMs or VLMs yourself with SFT and at least one RL method (GRPO, PPO, DPO, or similar), and you can explain what your reward or data changes did to the eval.

Nice-to-haves

• You understand computer vision fundamentals: detection and segmentation metrics, and the common architectures.

• 100% employer-paid premiums for employee coverage (medical, dental, vision)
• 401(k) with company match (100% of the first 3%, 50% of the next 2%)
• Meaningful equity with a 10-year post-termination exercise window
• Unlimited PTO

The range listed is base compensation. We offer participation in our bonus and variable compensation plans to select roles.