Website:
lh2.ai
Job details:
Job Title: Founding Engineering Lead
Location: On-site, Bengaluru
Employment Type: Full-Time
About LH2 AI Labs
Built by second-time founders who have built and sold companies before, LH2 AI Labs is building the post-training infrastructure for frontier AI models.
We bring private, high-quality institutional datasets and vetted domain experts into frontier AI pipelines across verticals such as coding, computer use, agentic workflows, medical, audio, and more.
For AI to keep progressing, it needs high-quality training data drawn from real production use cases. The public web has already been crawled and trained on - there is limited new signal left there. That is where we come in.
Our vision is to create a world where frontier models can access high-quality data on tap, the same way they access compute today.
About the role: Own the entire technical system that turns raw institutional data into lab-grade product - across three pillars: the data-to-delivery pipeline, the task-building infrastructure, and the RL environments that run those tasks - and lead the engineering team building all three.
The three pillars you own
Pillar 1 - Raw data to delivery-ready product. The ingestion pipelines that pull data from many enterprise sources, the PII-removal and cleaning layer, and the conversion and delivery of finished, machine-trainable product to labs and to other RL-environment companies. This is the business today; it has to be reliable, scalable, and privacy-safe.
Pillar 2 - Task-building infrastructure. Create, or help create and assemble, the infrastructure required to build tasks at scale — the mining/extraction pipelines, verifier and grader tooling, packaging, and the harness that turns raw substrate into validated tasks across verticals (coding first, then company ops, medical).
Pillar 3 - Environments. Stitch together the RL environments and run these tasks on them — reproducible sandboxes, the execution and orchestration layer, and the eval/rollout harness that runs tasks at volume to produce training data.
What you'll do across all three
- Architect each pillar and, critically, the shared infrastructure beneath them (sandboxing, reproducibility, orchestration, data handling) so the three reinforce each other instead of being rebuilt three times.
- Set engineering standards, patterns, and the technical roadmap; lead and grow the existing engineering team across the pillars.
- Make the build-vs-defer calls that keep a small team shipping — what to centralize, what's reusable across verticals, what stays modular for future verticals.
- Own reliability and reproducibility — flaky pipelines and non-deterministic environments are worthless; determinism is the job, especially in Pillars 2 and 3.
- Architect for the trust boundary in Pillar 1: sensitive data handled safely, often inside customer environments, with provable de-identification.
- Partner with Research (who defines task/verifier quality in Pillars 2–3) and with the Product Manager (who owns the Pillar 1 buyer-facing product).
Must-have:
- Bachelor’s or master’s degree from a leading Tier 1 institution
- 7+ years engineering, including senior/lead ownership of production data or ML infrastructure — you've built and operated real systems, not just designed them.
- Strong in data pipelines (ingestion, ETL/ELT, multi-source, incremental sync) and/or ML/serving infrastructure.
- Deep comfort with containerization, orchestration, and building reproducible, sandboxed environments at scale.
- Ability to set architecture and patterns and lead engineers in a small, fast founding team.
- Judgment about what to build once and reuse vs. what to leave modular per vertical.
Nice-to-have:
- Experience with RL environments, agent sandboxes, eval harnesses, or coding-task infrastructure (SWE-bench-style, Terminal-Bench-style).
- Privacy/PII-aware data processing or de-identification systems.
- Experience deploying inside customer VPC / private-cloud environments.
- Familiarity with the post-training data stack and what labs consume.
Click on Apply to know more.
