AI/ML Engineer
I'm an AI engineer specializing in building production-grade systems with large language models, multi-agent architectures, and scalable infrastructure. With a background in research and deployment, I bridge cutting-edge AI with real-world impact—scaling solutions to 150K+ users, reducing hallucinations by 40%, and winning hackathons for autonomous agent platforms.
I thrive on solving complex problems, from architecting RAG pipelines to ensuring AI safety and reliability in high-stakes environments.
- LLMs & Agents: Multi-agent orchestration (LangGraph, LangChain), fine-tuning (Llama, Mistral), RAG systems, prompt engineering, and evaluation frameworks for reasoning and hallucination mitigation.
- AI Infrastructure: AWS (SageMaker, Kubernetes, CloudWatch), Docker, CI/CD, MLOps for 99%+ uptime deployments.
- Research & Analysis: Graph neural networks (GNNs like GraphSAGE, GAT), anomaly detection, AI safety (alignment, oversight), and empirical evaluations on 18K+ test cases.
- Full-Stack Integration: Python/FastAPI backends, Redis/Qdrant for state management, WebSockets for real-time apps.
Languages & Frameworks: Python (PyTorch, Hugging Face), SQL, FastAPI, LangChain/LangGraph
Data & ML: Pandas, NumPy, Scikit-learn, Transformers, Weights & Biases
Cloud & Tools: AWS/GCP, Docker, Kubernetes, Git, Redis, Qdrant
Other: GraphQL, CUDA/GPU optimization, A/B testing
- Advancing responsible AI through Perplexity fellowship: Governance, bias mitigation, and scalable oversight.
- Exploring mechanistic interpretability and weak-to-strong generalization for safer AI systems.
- Building decentralized agent platforms for enterprise automation.
- 1st Place, Epiminds Multi-Agent Hackathon (2025)
- Valedictorian, University of Birmingham MSc (2024)
- Employee of the Year, Inzeitech (2024)
Email: tirthkanani18@gmail.com
LinkedIn: linkedin.com/in/tirthkanani
Website: tirthkanani.com
GitHub: tirth8205
Building reliable AI that scales—always iterating.