--- title: 'Member of Technical Staff, Post-Training at Chakra Labs' canonical: 'https://feeny.ai/job/member-of-technical-staff-post-training-chakra-labs-brooklyn-6dgqnwep2pcp' type: 'job' last_seen: '2026-09-11' --- # Member of Technical Staff, Post-Training at Chakra Labs - **Company:** Chakra Labs - **Location:** Brooklyn, NY - **Employment:** full-time - **Work type:** onsite - **Posted:** 2025-05-06 - **Last confirmed live:** 2026-09-11 - **Apply:** https://jobs.ashbyhq.com/chakra-labs/07ef4897-cb38-4eda-856b-71068ebeeea8 ## Job description ## About Us Chakra Labs' mission is to encode human taste into intelligence. We build high-fidelity environments, evals, and datasets for frontier AI research, working with several of the top labs. Our work sits at the frontier of post-training, agent environments, data quality, and research infrastructure. We care about building systems that make models better in ways that are measurable, useful, and hard to fake. ## What You’d Work On - Post-training loops. You’d help design and run model improvement workflows across supervised fine-tuning, preference optimization, and reinforcement learning approaches like GRPO. The work is not just launching training jobs; it’s figuring out what signal matters, how to collect it, and whether the model actually improved. - Environment and task design. We build environments that feel real and scenarios that push agents past static benchmark behavior. You’d design tasks, tools, validators, reward signals, and evaluation harnesses that test meaningful capabilities instead of whatever is easiest to measure. - High-fidelity trajectories. You’d create, inspect, and improve the data that teaches models how to behave. That means caring about taste, correctness, edge cases, and whether a trajectory would actually help a frontier model learn. - Reward and evaluation systems. You’d work on reward functions, rubrics, validators, and analysis tools that turn messy model behavior into useful training signal. You should be interested in where evals lie, where rewards get hacked, and how to make measurements more robust. - Training and research infrastructure. You’d run experiments across distributed GPU clusters, work with PyTorch and FSDP, and build the infrastructure needed to support model training, evaluation, and data generation at scale. - Customer research problems. You’d work with frontier AI labs to translate ambiguous research goals into concrete environments, datasets, experiments, and deliverables. ## About You - Machine learning fundamentals. You have Masters / PhD-level knowledge of machine learning fundamentals. You understand linear algebra, optimization, stochastic gradient descent, and can reason from first principles when a model or training run behaves unexpectedly. - Hands-on post-training experience. You have worked with or deeply understand LLM fine-tuning, preference data, reward modeling, or reinforcement learning for language models. You know the difference between reproducing a recipe and understanding why it works. - Environment-building instinct. You are excited by agent environments, multi-turn tool use, sandboxed tasks, eval harnesses, and the question of how to test capabilities that do not fit neatly into a benchmark. - Python and PyTorch fluency. You are strong in Python and comfortable building with PyTorch, FastAPI, and modern ML infrastructure. You can work at a high level, but you are also comfortable dropping into lower-level primitives when the abstraction leaks. - Strong engineering judgment. You can move between research ambiguity and production constraints. You know when to iterate quickly, when to be rigorous, and when a result is too fragile to trust. - Experience. No hard rule. Roughly 3-5 years is what we imagine, but more or less experience works if the expectations above resonate with you. What Makes This Different - The work is concrete. You are not just “touching the latest AI stack.” You are building the environments, trajectories, evals, rewards, and training loops that frontier labs use to improve models. - It is research-facing, but production-minded. Our customers are AI researchers and labs pushing the edge of what agents can do. The systems you build need to support real experiments, real users, and real deadlines. - Ownership, not theater. You own whole problems, not isolated tickets. One week you might be designing a new environment; the next you might be debugging a training run, improving a reward function, or scaling an eval pipeline. - Ambiguity is part of the job. There is no fixed playbook for this work. Data, post-training, and agent evaluation are changing quickly. If you need every problem to be fully specified before you begin, this role will be challenging. - The team. Our team is ex-Stripe, Snap, AWS, Microsoft, Airtable — you'll work with a small team who has years of shipping high-impact products over the last decade. ## About Chakra Labs ## Company Overview - **One-liner**: Chakra Labs builds deterministic, pixel-perfect reinforcement learning environments and high-fidelity trajectory datasets to push the boundaries of frontier AI agent research. - **Entity Type**: Private (Venture-backed; $10.1M total funding) - **Headquarters**: Brooklyn, New York, United States (with presence in Australia and Nigeria) - **Founded**: Not publicly available (recent; rapid growth from ~4 to 15 employees in past year) - **Founders**: Nirmal Krishnan (Co-Founder, CEO), Alexander Fung (Co-Founder, CTO) ## Core Business - **Primary industry**: Data Infrastructure and Analytics / Frontier AI Research Infrastructure - **Target customers**: Frontier AI research teams, model developers, and applied research labs (B2B / Research) - **Mission or purpose**: To provide research-grade infrastructure for frontier-scale experiments where emergent behaviors develop, enabling reproducible and scientifically rigorous agent development. ## Products & Services - **[Dojo](https://www.chakra.dev/)**: A collaborative reinforcement learning (RL) environment suite for computer-use agents (CUAs). Supports frame-accurate state capture, deterministic execution, mixed-modality training (GUI + MCP tool-use), and bespoke task generation. Available via containerization for hosted or on-premises deployments. - **Trajectory Datasets**: Curated, high-fidelity datasets (2,500+ hours of trajectories) preserving temporal structure and decision trees, designed for post-training and evaluation. Access via request. - **GLADOS-1**: A computer-use agent model post-trained using crowd-sourced trajectories (announced Sep 2025). ## Market Standing - **Valuation**: Not disclosed - **Key Metric**: Total Funding – $10.1M (Venture Round, January 2026) - **Notable Investors/Partners**: Not explicitly named in available sources; collaborated with "leading research teams" - **Growth Signals**: 366.7% headcount growth year-over-year (from ~4 to 15 employees); strong traffic growth (+88.7% monthly visits); publications and thought leadership in RL environment design. ## Competitive Advantages - **Deterministic, pixel-perfect environments**: Frame-accurate state control and temporal integrity prevent reward hacking and ensure reproducible experiments. - **Mixed-modality training**: Simultaneous GUI interaction and programmatic tool-use within single tasks – a differentiator for multi-modal agent research. - **Bespoke task generation**: Automated creation of novel, verifiable tasks enabling experiments with extended autonomous runtime and error recovery. - **High-fidelity datasets**: Realistic, in-distribution trajectories that avoid synthetic edge cases, improving transfer learning to deployment. ## Strategic Focus - Building infrastructure for frontier-defining agent problems, with emphasis on scientific rigor, speed of iteration, and collaboration between humans and AI. - Expanding the Dojo platform and dataset offerings to support both hosted and on-premises deployments for leading research labs. ## Why Work Here - **Culture**: Small, high-impact team (15 people) composed of talent from Stripe, Jane Street, Alchemy, The Boring Company, and UNSW CompClub – emphasis on engineering excellence and research. - **Remote-first**: Employees work remotely with offices listed in the US (Brooklyn) and presence in Australia and Nigeria. - **Research environment**: Focus on open platform for RL environments, publications, and pushing capability boundaries – ideal for engineers and researchers passionate about AI agent infrastructure. - **Tech stack**: Modern tooling (notion for internal ops, containerized deployment via Harbor and Verl). ## Sources 1. [chakra.dev](https://www.chakra.dev/) 2. [jobs.ashbyhq.com/chakra-labs](https://jobs.ashbyhq.com/chakra-labs) 3. [linkedin.com/company/chakra-labs](https://www.linkedin.com/company/chakra-labs) 4. [builtin.com/company/chakra-labs](https://builtin.com/company/chakra-labs) 5. [chakra.dev/methodology](https://www.chakra.dev/methodology) ## Other roles at Chakra Labs - [Strategic Projects Lead](https://feeny.ai/job/strategic-projects-lead-chakra-labs-brooklyn-52rvmpj8xb5b) — Brooklyn, NY - [Member of Technical Staff, Forward Deployed](https://feeny.ai/job/member-of-technical-staff-forward-deployed-chakra-labs-brooklyn-pj8acx7ee2j3) — Brooklyn, NY - [Member of Technical Staff](https://feeny.ai/job/member-of-technical-staff-chakra-labs-brooklyn-sxmmc376tx54) — Brooklyn, NY - [Member of Technical Staff, Post-Training](https://feeny.ai/job/member-of-technical-staff-post-training-radical-numerics-san-francisco-e3h9ssf1mhqk) — San Francisco, CA - [Member of Technical Staff, Post-Training](https://feeny.ai/job/member-of-technical-staff-post-training-cohere-london-8efxepvqegr8) — London, United Kingdom