Looking for a job

Ayush Kumar

@ayushkumar18

Message

ML-focused software engineer building scalable, auditable retrieval and backend systems.

India

Message

What I'm looking for

I seek an internship or early-career role building production ML systems that combine retrieval, deep learning, and resilient backend engineering, with opportunities for research impact and system-level responsibility.

I am an ML-focused software engineer experienced in building scalable, production-ready systems that combine retrieval, deep learning, and robust backend services. I design auditable pipelines and deploy services that emphasize reproducibility, low latency, and correctness under failure modes.

My recent work includes an agentic RAG pipeline for finance documents with layout-aware parsing, hybrid search using pgvector and BM25, and a FastAPI service with Redis caching and index-versioned invalidation. I also engineered fault-tolerant inventory and order management services using Node.js, PostgreSQL, Redis, Docker, and AWS, emphasizing transactional correctness and idempotent request handling.

I have research experience in spatial deep learning under weak supervision, producing significant gains through self-supervised pretraining and rigorous spatial cross-validation, with a paper under review at IJCNN 2026. I hold awards and certifications including a Best Paper (ICAIO 2025) and several ServiceNow and AWS certifications.

Experience

Work history, roles, and key accomplishments

Current

ML/AI Intern

Current

Doctoral Institute for Evidence Based Policy

Oct 2025 - Present (4 months)

Built a finance-focused document intelligence pipeline over SEC filings and transcripts, implemented layout-aware PDF parsing and provenance, and deployed a FastAPI RAG service with caching and agentic query planning to produce citation-grounded answers.

Python fastAPI PostgreSQL Pgvector Redis RAG PDF Parsing

Research Intern

Indian Institute of Technology Bombay

Apr 2025 - Jan 2026 (9 months)

Developed end-to-end deep learning pipelines for large-scale spatial prediction under weak supervision, engineered spatial datasets and validation, and improved average precision by 21% with regularization and self-supervised pretraining.

PyTorch Deep Learning Positive–Unlabeled Learning Data Engineering Cross Validation

Software Engineer (Project)

Tata Steel

Apr 2024 - Dec 2024 (8 months)

Designed and implemented a resilient inventory and order management backend using Node.js, PostgreSQL, Redis and Docker, ensuring transactional correctness, idempotency, and fault tolerance under concurrent updates and dependency outages.

Node PostgreSQL Redis Docker AWS

Education

Degrees, certifications, and relevant coursework

SRM Institute of Science and Technology

Bachelor of Technology, Computer Science and Engineering

2023 -

Activities and societies: Relevant coursework: Machine Learning, Database Management Systems (DBMS).

Pursuing a Bachelor of Technology in Computer Science and Engineering with coursework in Machine Learning and Database Management Systems; expected graduation in 2027.