Generative AI / Agentic AI Engineer with 3+ years at Accenture, shipping production AI for Fortune 10/100 clients across healthcare, aviation, and entertainment.
- 🤖 Designed a multi-agent failure-triage system that cut resolution time from 1–2 days → under 10 minutes
- 🏥 Architected a real-time healthcare call assistant processing 5,000+ live utterances/day with sub-second latency and 99.5% uptime
- 🎬 Delivered a GPT-4o document automation suite for a Fortune 100 entertainment company — 70–80% effort reduction across 15+ departments
- 🧪 Built 8 open-source GenAI projects with 2,400+ tests spanning the full GenAI stack
- 📍 Based in Noida, India — open to GenAI / Agentic AI opportunities
| # | Project | What it does | Key Tech | Tests |
|---|---|---|---|---|
| 1 | DocuMind | Production-grade RAG — PDF ingestion, hybrid search, Cohere reranking, cited multi-LLM answers & RAGAS evaluation | RAG · BM25 · Reranking · FastAPI | 9 |
| 2 | AgentForge | Multi-agent AI framework with tool-using agents, ReAct reasoning, orchestration patterns & Streamlit dashboard | ReAct · Tool Use · LangGraph · CrewAI | 62 |
| 3 | ModelForge | LLM fine-tuning pipeline — QLoRA on Mistral 7B with BLEU/ROUGE/BERTScore eval, LLM-as-Judge & cost analysis | QLoRA · PEFT · Mistral 7B · HuggingFace | 577 |
| 4 | GuardianAI | LLM safety & observability — content filtering, PII detection, token/cost tracking & real-time dashboard | Guardrails · PII · Drift · Streamlit | 457 |
| 5 | ShipAI | ML deployment pipeline — Docker, FastAPI serving, 3-tier health checks, GitHub Actions CI/CD & load testing | Docker · FastAPI · CI/CD · Locust | 650 |
| 6 | VisionRAG | Multimodal RAG — PDF + image ingestion with OCR, vision embeddings, hybrid retrieval & multimodal generation | OCR · Vision · Multimodal · RAG | 293 |
| 7 | InterviewGPT | AI mock interview system — adaptive difficulty, 14 heuristic scoring functions & progress analytics | Adaptive Engine · Heuristics · OpenAI | 295 |
| 8 | ToolBridge | Prompt engineering toolkit — template management, chain-of-thought, few-shot optimization, A/B testing & versioning | Prompt Eng · A/B Testing · Versioning | 95 |
📊 Total: 2,438 tests across 8 production-grade projects
| LLM & RAG | OpenAI GPT-4o · Anthropic Claude · Mistral · LangChain · LangGraph · CrewAI · Pinecone · ChromaDB · Azure AI Search · BM25 · Hybrid Search · Cohere Reranking |
| Agentic AI | Semantic Kernel · Azure AI Foundry · Tool Calling · Function Calling · MCP Servers · ReAct Pattern · Human-in-the-Loop · Multi-Agent Orchestration |
| Fine-Tuning | QLoRA · LoRA · PEFT · RLHF · SFTTrainer · HuggingFace Transformers · Weights & Biases |
| Evaluation | RAGAS · BLEU · ROUGE · BERTScore · LLM-as-Judge |
| Safety & MLOps | Guardrails · PII Detection · Prompt Injection Prevention · Drift Detection · A/B Testing · LLMOps |
| Cloud & DevOps | Azure (OpenAI, Functions, Event Hub, Cosmos DB, App Service, Key Vault) · Docker · GitHub Actions · Azure DevOps |
| Backend | Python · Flask · FastAPI · Pydantic · pytest · REST · Microservices · Event-Driven Architecture |
| 🌟 Sparkling Star Award 2024 · Accenture & Avanade |
Recognized for building a multi-agent GenAI system adopted across enterprise workflows |
| 🏆 ACE Award (Team) 2023 · Accenture |
Awarded for the real-time healthcare AI assistant — 5,000+ daily utterances, sub-second latency, 99.5% uptime |
GenAI Portfolio · 8 Projects · 2,400+ Tests
┌────────────────────────────────────────────────────────────────┐
│ │
│ 📄 DocuMind 🤖 AgentForge 🧬 ModelForge │
│ Production RAG Multi-Agent AI LLM Fine-Tuning │
│ │
│ 🛡️ GuardianAI 🚀 ShipAI 👁️ VisionRAG │
│ Safety & Observ. ML Deployment Multimodal RAG │
│ │
│ 🎯 InterviewGPT 🔧 ToolBridge │
│ Mock Interviews Prompt Engineering │
│ │
└────────────────────────────────────────────────────────────────┘
▲
│
┌───────────────┴────────────────┐
│ Live Portfolio Site (GH Pages) │
│ aditya2425.github.io/Portfolio │
└────────────────────────────────┘
⭐ If you find any of these useful, a star goes a long way.
Open to GenAI / Agentic AI roles · Reach out: aditya78332412@gmail.com