Applied AI R&D Engineer

Testsigma is an AI-native company building autonomous agents that verify software as fast as AI can write it.

AI has transformed how code is written, but not how it is trusted. When the same AI writes both a feature and the checks for it, both inherit the same blind spots. Our agentic AI coworkers act as the independent judge: they understand what changed in an application, identify what is at risk, adapt as the application evolves, and tell engineering teams whether a release is safe to ship, bringing in a human whenever judgment is needed.

We're solving this in production for global enterprises such as Cisco, Samsung, DHL and IndusInd Bank. Testsigma is headquartered in San Francisco, with engineering in Bangalore, and is backed by MassMutual, Accel, Strive and Bold Cap.

THE ROLE

You'll join our AI team building the core agent layer of our platform, working directly with the founding team. The team is small, so your architecture decisions will shape the platform for years.

This isn't a role for someone who chains LLM API calls and calls it agentic. The systems you build have deterministic rule layers, multi-stage reasoning pipelines, confidence scoring when the answer is ambiguous, feedback loops that improve over time, and escalation logic that knows when to stop and ask a human.

WHAT YOU'LL DO

• Build multi-agent orchestration: agents that plan, execute, correct themselves and escalate, owning the intelligence behind our autonomous agents for generation, self-repair, judgement and gap detection

• Build context graph infrastructure: a living knowledge graph connecting requirements, DOM elements, agent-generated checks, code functions and user behaviour signals

• Build browser crawl agents that understand a running application by meaning, not just structure

• Build the agent evaluation layer that tells us whether an agent's decision was correct, and gates every prompt and model change

• Integrate LLMs at production scale: prompt engineering, structured output, context window management, fallback strategies and confidence calibration

• Build feedback loops that make the system measurably smarter with every agent run, self-repaired step and confirmed bug

MUST-HAVE

• 3+ years of software engineering, including recent hands-on work building multi-agent systems in production, not prototypes or demos

• Hands-on experience with LangGraph, LangChain, CrewAI, AutoGen or an equivalent orchestration framework, and the ability to explain why you chose it

• Designed and queried knowledge graphs or graph databases, such as Neo4j or a graph layer on a relational system

• Built systems that detect absence: not just what's wrong, but what's missing

• Production Python that is async, typed, modular and observable

• Browser-use or equivalent browser and computer-use agents, beyond basic scripting

GOOD TO HAVE

• Agent evaluation frameworks, such as RAGAS, custom evals or LLM-as-judge pipelines

• Vector stores alongside graph databases (pgvector, Pinecone, Weaviate), and knowing when to use which

• Developer tooling or AI infrastructure products

• GitHub API, Jira API or similar developer integrations

• Some TypeScript or Node.js

WHAT WE OFFER

• Direct access to the founding team and real influence over product direction

• The chance to be a core architect of a category-defining platform

• 100% company-paid medical insurance

• Daily catered lunch and free snacks