Research Engineer

Tessera Labs•San Jose, CA

4d•$200,000 - $300,000

About The Position

Tessera Labs is developing an AI platform designed to transform how large enterprises manage their decades of accumulated processes, data, and code. Unlike traditional approaches that are lengthy and costly, Tessera acts as a transformation engine, understanding and modifying enterprise systems as a connected whole within weeks. The platform is vendor-agnostic, working with systems like SAP, Salesforce, Workday, Oracle, Snowflake, and MuleSoft. Key challenges include ensuring governance (traceability and reversibility of actions) and generality (applicability to diverse and complex enterprise landscapes). Tessera Labs sells a product, not a service, and has secured $60M in Series A funding from investors including Andreessen Horowitz.

Requirements

Significant experience training, fine-tuning, or post-training language models, with demonstrable results.
RL tuning experience (RLHF, RLAIF, RLVR, GRPO-family, or agentic RL) is nearly a requirement.
Experience with memory and context for long-running agents (architecture, retrieval, or training).
Strong software engineering fundamentals; experimental code should be runnable by others.
Fluent in Python and PyTorch (or JAX).
Comfortable debugging distributed training.
Ability to design, run, and interpret experiments with empirical rigor, distinguishing effects from noise and bugs.
Experience with GPU infrastructure at scale, understanding time and memory usage.
Desire for research to be productionized, treating it as a constraint rather than a burden.
Clear written communication skills for decision-making documents.

Nice To Haves

Experience building RL environments, execution sandboxes, or verifiable-reward task suites.
Experience with long-context modeling (extension, efficient attention, position methods, or evaluation).
Experience with knowledge graphs, ontologies, or semantic layers over structured enterprise data.
Experience with agent memory systems (episodic or otherwise) in production.
Experience with code models (repository-scale context, program synthesis, automated repair, or transpilation).
Contributions to open-source ML systems (vLLM, SGLang, PyTorch, Triton, DeepSpeed, Ray, Megatron, TRL, or similar).
A track record of publications, technical reports, or open-source releases.
An advanced degree in CS, ML, math, physics, or a related quantitative field, or equivalent industry research experience.

Responsibilities

Build and scale the post-training stack: SFT, preference optimization, and reinforcement learning for long-horizon tool use, transformation, and reconciliation over enterprise systems.
Build the memory and context machinery that long-horizon agents run on, including retention, structure, retrieval, compaction, and revision.
Build the representation layer that agents reason over, such as ontologies and knowledge graphs derived from enterprise systems, and the pipelines to construct, validate, and maintain them.
Design and implement data generation and curation pipelines (synthetic landscapes, transformation traces, tool-call trajectories, curriculum infrastructure) to train models on systems not seen in public.
Build RL environments: sandboxed landscapes and execution-and-verification harnesses for automated change application, checking, and scoring, utilizing design-partner traces where synthetic data is insufficient.
Build and run the offline eval harness for long-horizon agentic behavior, including trajectory-level scoring, task suites, and reproducible result infrastructure.
Run experiments end to end: design, launch, debug, and analyze, distinguishing real effects from noise.
Optimize training and inference throughput, focusing on kernels, parallelism strategies, memory, batching, and serving, especially for long contexts.
Take a training result from 'the eval moved' to 'it's serving traffic', including quantization, serving configuration, and rollback paths.
Establish standards for reproducibility, experiment tracking, and result hygiene.