Ashutosh Yadav_

$ echo "full stack engineer"

AI engineer with 2+ years bridging applied AI and enterprise software — production RAG systems and LLM workflows in Python/LangGraph, resilient backends in Spring Boot. Currently completing an MSc in Computer Science at University College Dublin.

open_to_work: true · ai / software engineering roles

./view_projects --download_cv

~/assets/profile.png

// fig 1. the human behind the commits

$ cat ~/about.md

about_me

~/about.md

I'm an AI engineer who likes owning problems end to end — production RAG systems and agentic LLM workflows in Python and LangGraph, resilient services in Java Spring Boot, and the cloud plumbing that keeps them honest. Currently completing an MSc in Computer Science at University College Dublin.

In industry I've shipped AI platforms at EXL Service, enterprise SaaS used by Sony and Universal Studios at Capgemini, and private-equity data pipelines at Morningstar. I care about systems that are fast, observable, and boring to operate — in the best possible way.

linkedin github email

~/stack.json

// technologies I reach for, grouped by layer

{
  "languages": ["java", "python", "typescript", "sql"],
  "ai_genai":  ["rag", "langgraph", "bedrock", "mcp"],
  "backend":   ["spring_boot", "fastapi", "node", "jwt"],
  "frontend":  ["react", "angular", "websocket"],
  "devops":    ["docker", "kubernetes", "helm", "ci_cd"],
  "databases": ["pgvector", "neo4j", "mongodb", "faiss"]
}

$ git log --graph --all

career.log

commit e5b7d20 (HEAD → main, tag: exl-service)

AI Software Engineering Intern

EXL Service · Dublin, Ireland

EXL

Date: Jun 2026 — Present

Independently shipped a production-grade Agentic RAG platform within two weeks of joining — S3 knowledge-base sync via LiteLLM, pgvector storage with fine-tuned chunking, de-duplication, and a LangFuse LLM-as-judge evaluation pipeline
Migrated the RAG agent's knowledge base to AWS Bedrock with automated document sync and health checks — containerised service deployed to Dev/QA via Docker and Helm
Built a graph-based fraud-detection agent over Neo4j (45M+ transactions), combining LangGraph tool-calling with a supervised ML risk-scoring model tuned for high recall
Extended the platform MCP server with tools for an internal chatbot to query and edit AI, agent, and model metadata — adding chat-history context and Bedrock Guardrails

commit a3f92c1 (tag: capgemini)

Software Engineer

Capgemini Technology Services · Mumbai, India

Date: Mar 2024 — Aug 2025

Developed and maintained full-stack enterprise features — Java, Spring Boot, Maven and REST APIs on the backend, JavaScript/jQuery on the frontend — shipping 5+ production features across bi-weekly Agile Scrum releases for Sony & Universal Studios
Built event-driven workflows with Java EJB and asynchronous processing — reliably handling 1,000+ weekly automated transactions with validated data integrity via SQL and JUnit
Deployed builds to remote Linux servers over SSH across DEV, staging, and UAT — reducing release cycle times by 30%

commit 7d41be0 (tag: morningstar)

Data Analyst Associate

Morningstar, Inc. · Mumbai, India

Date: Oct 2023 — Feb 2024

Automated data processing and validation pipelines for large-scale private-equity datasets — improving workflow efficiency by 30%
Delivered analytical insights that contributed to a 15% increase in client retention

$ ls -la ~/projects

projects/

~/projects/snapsearch

snapsearch

Privacy-first macOS desktop app for natural-language image search over large photo libraries — a 3-stage on-device pipeline (Florence-2 captioning, Sentence Transformers embeddings, FAISS HNSW retrieval) with 95%+ accuracy and Apple Silicon MPS acceleration.

python
pytorch
florence-2
faiss

open_gallery() →

~/projects/inflation-dynamics

inflation-dynamics

Econometric analysis of Malaysian inflation volatility (1990–2006): backtested GARCH(1,1) models, identified structural breaks, quantified policy-shock impact.

python
statsmodels
arima

read_report.pdf →

~/projects/aetherwings

aetherwings

Microservices flight-scheduling system — RabbitMQ for async messaging (25% lower latency) with JWT-secured services.

java
spring
microservices

open_gallery() →

~/projects/soundsphere

soundsphere

Full-stack music-streaming platform with dynamic playlist management, RESTful APIs, and a responsive React UI.

node
react
mongodb

open_gallery() →

~/projects/skillhub

skillhub

Modular LMS built on Java microservices with an API Gateway — Spring AI powers a conversational training assistant, with JWT-based RBAC across Admin, Trainer, and Learner roles.

java
spring_ai
angular
jwt

open_gallery() →

~/projects/finance-copilot

$ docker build -t finance-copilot .

$ gcloud run deploy finance-copilot \

--region europe-west1 --allow-unauthenticated

✓ deployed · revision ready · traffic 100%

> ws: streaming retrieval 0.42s · 12 chunks

> rag: hnsw index warm · latency ok

finance-copilot

Containerised FastAPI/React app on GCP Cloud Run with automated CI/CD — a RAG pipeline with HNSW vector indexing and real-time WebSocket streaming for sub-second document retrieval.

fastapi
react
typescript
gcp

open_live_app →

~/projects/predire

predire

Disease-prediction engine using Random Forest classifiers over symptom data to surface likely health conditions.

python
sklearn
ml

open_gallery() →

$ git shortlog -sn --community

open_source

~/oss/seed4j-cli

seed4j-cli npm

Enhanced the help command to surface example values for module parameters by surfacing defaultValue definitions from the core library — improving developer experience and CLI discoverability. Merged and live on npm.

java
cli
maven

view_on_npm →

~/oss/db2rest

db2rest github

Diagnosed and fixed a PostgreSQL boolean type-casting bug (repo with 457+ stars) where RSQL filters like active==true caused runtime type mismatches — patched BaseDialect to return native Java Boolean objects and added a regression test.

java
postgresql
junit

view_on_github →

$ cat ~/education.json

education

~/education/ucd.json

MSc Computer Science

University College Dublin · Dublin, Ireland

sept 2025 — sept 2026 · gpa: 3.40/4.20

in_progress

~/education/tcet.json

B.Eng Information Technology

Mumbai University · Thakur College of Engineering

may 2019 — jun 2023 · cgpa: 8.34/10

$ ls ~/certifications

azure_fundamentals (AZ-900) accenture_data_analytics java_masterclass sql (coursera)

$ open ~/publications

publications

~/publications/disease-prediction.md

“Disease Prediction System”

Published in the International Journal of Innovative Research in Technology (IJIRT) at the Multicon W conference.

read_paper.pdf

$ ssh ashutosh@dublin

get_in_touch

ashutosh@dublin: ~/contact

Currently looking for opportunities in AI engineering, quantitative development or software engineering. If that sounds like a fit — or you just want to talk shop — my inbox is open.

$ echo "ashutoshy115@gmail.com"

$ call "+353 899447863"

./send_email --open_linkedin