AI/ML Engineer

Hi, I'm
Aditya Agrawal

Building AI-powered systems and scalable backends that solve real-world problems. From RAG chatbots to anomaly detection — I craft intelligent solutions with code.

Aditya Agrawal AA

Aditya Agrawal

AI/ML · Software Engineer

Mumbai, India

Open to opportunities
40%
Tickets Reduced
90%
Faster Response
81%
Answer Accuracy

Building the future with AI & code

I'm a software engineer passionate about combining machine learning with robust backend systems to create impactful products.

I currently work at Accenture where I design and build AI-powered systems that automate supplier support using RAG architecture, LLMs, and semantic search. I've reduced support tickets by 40% and improved response times by 90%.

Previously, I built enterprise Angular applications at Impetus, creating reusable UI components and integrating REST APIs for document management systems used across departments.

I'm deeply interested in LLMs, vector retrieval, and microservices. I believe in writing clean, testable code and shipping with confidence.

2,000+
Global Users
15+
Reusable Components
20+
API Integrations
<500ms
Anomaly Detection

Where I've worked

Building impactful products at scale.

Software Engineer
Accenture
10/2024 – Present · Pune
  • Developed an AI-powered GS3 Supplier Support Chatbot using RAG architecture, LLMs, and semantic search to automate supplier query resolution.
  • Reduced repetitive Level-1 and Level-3 support tickets by 40% and improved query response times by 90% through intelligent automation.
  • Built REST APIs using Python & FastAPI, enabling real-time chatbot interactions for 2,000+ global supplier users.
  • Improved answer accuracy to 81% through vector embeddings, prompt engineering, and retrieval optimization.
RAGLLMsFastAPI LangChainVector Search Python
Angular Intern
Impetus Technology Pvt. Ltd.
06/2023 – 09/2023 · Indore
  • Developed key modules for an enterprise Document Management System (DMS) using Angular 16, supporting 1,000+ business documents across departments.
  • Designed and implemented 15+ reusable UI components using Angular, TypeScript, HTML5, CSS3 & Bootstrap, reducing development effort by 30%.
  • Integrated 20+ RESTful APIs for document upload, retrieval, version control, and approval workflows.
  • Implemented advanced search, filtering, and categorization, reducing document retrieval time by 40%.
  • Resolved 50+ UI and functional defects, improving application stability by 20%.
Angular 16TypeScriptREST APIs BootstrapHTML5/CSS3

Tech stack & expertise

Tools and technologies I work with daily.

AI / LLM

RAGLangChainLangChain4j OpenAIClaudeGemini OllamaQwenPrompt Engineering Function CallingAgentic Workflows Multi-Model Routing

Vector & Retrieval

FAISSChromaDBEmbeddings Semantic ChunkingHybrid Retrieval Isolation Forest

Backend

JavaPythonSpring Boot Spring SecurityFastAPI REST APIsMicroservices HibernateJPA

Data & Messaging

PostgreSQLMySQLMongoDB RedisApache Kafka

DevOps & Cloud

DockerKubernetesGitHub Actions JenkinsGCPNginx

Frontend & Tools

Angular 17+TypeScriptGit JUnitMockitoPostman Swagger/OpenAPIMavenGradle

Featured work

A selection of my projects — from AI to microservices.

AI-Driven Microservices Anomaly Detection

Built a Kafka/Spring Boot pipeline to ingest real-time microservice telemetry (latency, CPU, error rates). Deployed an Isolation Forest model that detects anomalies with <500ms end-to-end latency. Optimized inference to ~8.8ms via Redis feature caching; exposed anomaly scores via REST/API Gateway for dashboards and alerting.

Spring BootKafkaRedis Isolation ForestREST

LLM Cost Autopilot

Architected a multi-provider LLM routing platform (Groq, OpenAI, Anthropic, Ollama) with FastAPI and a scikit-learn classifier that auto-routes requests by complexity, cutting inference costs by up to 70%. Built an asynchronous LLM-as-judge verification pipeline with a self-improving feedback loop, achieving 87.8% classification accuracy.

PythonFastAPIscikit-learn SQLAlchemyDocker

GS3 Supplier Support Chatbot

Enterprise-grade chatbot using RAG architecture, LLMs (OpenAI, Claude), and semantic search. Integrated with FastAPI and deployed for 2,000+ global suppliers. Achieved 81% answer accuracy with optimized retrieval and prompt engineering.

RAGLLMsFastAPI LangChainVector DB

Where I learned

Madhav Institute Of Technology And Science

Bachelor of Technology · Mechanical Engineering · Gwalior, M.P.

2020 – 2024

Let's build something great

I'm always open to interesting conversations and opportunities.

Reach out directly

agrawalwadi45@gmail.com

+91 9179711777

Mumbai, Maharashtra

Typically reply within 24 hours.