Open to opportunities

Anushka Jain

@anushkajain3

Message

Full-stack engineer building AI-powered LLM pipelines and dependable systems for real-world workflows.

India

Message

What I'm looking for

I’m looking for a role where I can build AI-driven products end to end—owning evaluation, reliability, and system design—using modern stacks and shipping quickly with measurable impact.

I’m a Full-Stack Engineer and Co-Founder who builds end-to-end AI systems—from data ingestion to reliable inference—while obsessing over evaluation quality and production-grade reliability. I’ve designed pipelines that turn unstructured inputs into structured, decision-ready outputs.

At Gradonix, I architected an “OCR + LLM answer-scoring pipeline” covering document ingestion, OCR extraction, prompt design, LLM inference, and structured score output. I iterated on scoring prompts and evaluation criteria to reach 94.3% scoring agreement with human evaluators across 300+ answer sheets, while cutting grading time by 60%.

Across my projects, I focus on safe execution under real constraints: I built an AI agent system for expense-claim approvals with crash-safe pause/resume and exactly-once execution using a LangGraph agent, Postgres checkpointing, and Redis-backed idempotency. I also delivered a pgvector-based resume-parsing and ranking pipeline, and a FastAPI search typeahead backend with consistent-hashing caching and strong query performance.

Experience

Work history, roles, and key accomplishments

Current

President, Student Council

Current

Student Council

Jan 2026 - Present (6 months)

Served as President of the Student Council, representing the student body to institute management. Raised and followed up on 1,200+ student-reported issues for resolution.

Stakeholder Management Student Advocacy Issue Tracking Leadership

ApprovalOps Project

ApprovalOps

Jan 2026 - Present (6 months)

Built and deployed an AI agent system to auto-approve or escalate expense claims using a LangGraph agent with Postgres-backed checkpointing for crash-safe pause/resume across restarts. Implemented an MCP server exposing policy/employee/history/duplicate detection/notification/execution/logging tools with two-layer idempotency for exactly-once behavior under concurrent requests.

fastAPI LangGraph PostgreSQL MCP Redis Gemini API Next.js AI Agents Tool calling

Search Typeahead System

Jan 2026 - Present (6 months)

Built a search-autocomplete backend using a consistent-hashing cache layer with replica-based distribution for minimal remapping on node changes. Implemented a batch write buffer that coalesces searches before flushing to SQLite, along with a time-windowed trending engine and B-tree indexed prefix search.

fastAPI SQLite Python Consistent Hashing Caching Prefix Search Trending Engine Benchmarking

Recruitify Project

Recruitify

Jan 2026 - Present (6 months)

Built a resume-parsing pipeline for an AI recruitment platform with asynchronous batch upload and OCR/parsing using pdfplumber. Fed parsed candidate data into a pgvector-based semantic ranking system and added a validation suite for pipeline quality.

Gemini fastAPI PostgreSQL Pgvector Pdfplumber Semantic Search Data Validation

Co-Founder & Full-Stack Engineer

Gradonix

Jan 2025 - Present (1 year 6 months)

Architected and built an OCR + LLM answer-scoring pipeline covering document ingestion, OCR extraction, prompt design, LLM inference, and structured score output. Improved scoring agreement to 94.3% versus human evaluators while cutting grading time by 60%.

LLM Prompt Engineering LLM Inference Evaluation Structured Outputs