Senior AI Engineer

Touchstones; India; Remote; INR 3,300,000 - 5,500,000; Full time

Benefits offered by this job

Equity

Remote work from India

Job summary

Touchstones, an AI-native hiring company, seeks a mid-senior engineer to build LLM-powered evaluation pipelines for role-specific work samples. You’ll work directly with the founder while shaping product decisions in a fully remote role from India.

Requirements include 5+ years in software engineering with 2+ years shipping LLM features, strong Python and/or TypeScript, experience with LLM APIs, structured outputs and RAG, plus clear written communication.

Qualifications

• 5+ years in software engineering with 2+ years shipping LLM features.
• Strong Python and/or TypeScript with experience in LLM APIs, structured outputs and RAG.
• Hands-on experience building evals and judging model quality with data.
• Clear written communication and ability to thrive in a startup environment.

Responsibilities

• Build LLM pipelines to evaluate work against rubrics with evidence citations.
• Design evaluation harnesses (golden sets, regression evals, recall/precision on evidence retrieval).
• Author and harden AI-engineering assessments: retrieval evaluation, tool-call guardrails, prompt-injection defenses.
• Ship reviewer features: summarize work trails, flag gaps, draft interview questions.

Skills

Software engineering

LLM features

Python/TypeScript

RAG/Tool calls

Communication

Tools

PostgreSQL

Next.js

Supabase

Job description

Touchstones is an AI-native hiring company. We build role-specific engineering work samples where candidates are allowed to use AI. We capture the work trail and map every piece of evidence to the hiring team's rubric, and a human reviewer makes the call. We've run 850+ candidate screens, growing 145% month over month, and we're backed by Mucker Capital. You'd be one of the first engineers, working directly with the founder.

Location: Remote (India). Full-time, Mid-Senior level.

What you'll do

• Build the LLM pipelines that evaluate captured work against each rubric criterion, with evidence citations and calibrated reasoning.
• Design evaluation harnesses (golden sets, regression evals, recall/precision on evidence retrieval) so scoring quality is measured, not assumed.
• Author and harden AI-engineering assessments: retrieval evaluation, agent tool-call guardrails, prompt-injection defenses.
• Ship agentic features for reviewers: summarizing work trails, flagging gaps, drafting follow-up interview questions.
• Own model choice, cost, latency and fallback across providers.
• Keep the product's principles intact: AI advises, humans decide, no biometric signals.

Requirements

• 5+ years in software engineering, including 2+ years shipping LLM-powered features to production.
• Strong Python and/or TypeScript; experience with LLM APIs, structured outputs, tool calling and RAG.
• Hands-on experience building evals and judging model quality with data.
• A product mindset and comfort with startup ambiguity.
• Clear written communication; able to overlap with US Pacific mornings.

Nice to have

• Experience with hiring/HR tech, assessment design or psychometrics.
• Background in agent safety, guardrails or red-teaming.
• Postgres/Supabase or Next.js experience.
• Open-source or public writing on AI evaluation.

Compensation

• India: competitive, up to 55 LPA (higher for exceptional fit), plus early-employee equity.