GenAI Researcher (Bengaluru)

GenAI Researcher (Bengaluru)

19 Nov
|
Arcfront
|
Bengaluru

19 Nov

Arcfront

Bengaluru

About Arcfront Arcfront is a research-and-data partner to the labs building frontier models. We build RL environments, proprietary data and evals that find where a model lags and help it hill-climb. We work with several of the leading frontier labs and a growing set of neo-labs, and we are growing quick. We believe that models need to get better on real world work, and real world breakthroughs. Interestingly, most of that work is present in some shape and form inside companies. As long as we can fish that out in a way where we can make that information useful, we can be a strong enabler to the world getting better intelligence models. We are a team of researchers and engineers passionate about moving the world forward. The role You'll own how we measure what models can and can't do in a capability area, from what gets measured, to why that measurement is valid, to what should change because of the result. You'll design new evaluations from scratch, pressure-test our own numbers, and turn findings into decisions. What you'll own
• A measurement roadmap for one capability area (for example agentic tool use, long-horizon coding, reasoning or retrieval faithfulness): what to measure now, what to retire, and what to deliberately leave for later.
• Evaluation design from zero: construct validity, item difficulty and headroom, the power to separate the models that matter, and a contamination threat model written before the first item is authored.
• Agentic eval infrastructure: sandboxed environments, deterministic tool mocks, trajectory-level scoring and partial credit over long horizons.
• Judge validation:



human label collection, chance-corrected agreement, bias audits and clear rules on where an LLM judge is and isn't trusted.
• Adversarial pressure on our own numbers: hunting for reward hacking, shortcut solutions, grader exploits and prompt-format sensitivity.
• Influence and mentorship: you make the argument yourself, and you raise the methodological bar for the engineers around you. What we're looking for
• 4–6 years in ML research engineering, applied research or quantitative measurement where being wrong was expensive.
• You've shipped measurement that other people relied on: an eval suite, scoring harness or experimentation platform.
• Statistical depth: paired and clustered designs, power analysis, multiple-comparison control, calibration. You can show a team their study was underpowered, with the arithmetic.
• Systems ability: large-scale batched inference, cost and rate-limit management, containerised sandboxes, reproducibility across thousands of runs.
• Design taste: you can look at a proposed eval and quickly name how its score will stop meaning what it claims.
• Clear, persuasive writing. Your memos change decisions. Signals that stand out
• Built or substantially contributed to a public benchmark, eval harness or leaderboard.
• Published work on evaluation methodology, contamination or judge reliability.
• A background in psychometrics, econometrics, experimental design or causal inference applied to ML.
• A time you proved an exciting result was an artefact and convinced the team.

📌 GenAI Researcher (Bengaluru)
🏢 Arcfront
📍 Bengaluru

Reply to this offer

Impress this employer describing Your skills and abilities, fill out the form below and leave Your personal touch in the presentation letter.

Subscribe to this job alert:

Get the latest job offers by email for: genai researcher (bengaluru) / bengaluru

Subscribe to this job alert:

Get the latest job offers by email for: genai researcher (bengaluru) / bengaluru