Sailesh Dwivedy's Personal Website

Hello! I’m Sailesh Dwivedy, an AI/ML Engineer with 5+ years of experience developing, deploying, and scaling production-ready solutions in Generative AI, LLM fine-tuning, RAG, Agentic AI, audio ML, NLP, and Machine Learning. I bring a unique blend of research-driven innovation and industry execution, having built and deployed AI systems across healthcare, finance, and academic domains for Fortune 500 clients.

You can connect with me on LinkedIn

Industry & Research Experience

I specialize in LLM fine-tuning, Retrieval-Augmented Generation (RAG), Agentic AI, knowledge distillation, and on-device audio ML, combining practical development expertise with research-oriented rigor.

• Currently working as an AI/ML Applied Scientist at NeuroSpring, where I build on-device deep learning models for speech abnormality detection using PyTorch and Hugging Face. I also developed an Agentic RAG chatbot that enables physicians to query patient PDFs with grounded responses, reducing clinical triage time by 60%.

• As a Graduate AI/ML Engineer at the Image and Video Computing Lab, University of Colorado Boulder, I fine-tuned vision models and developed a novel knowledge distillation technique for computer vision. This layer selection strategy achieved 80% training efficiency and 15% accuracy gain over SOTA models, and was submitted to NeurIPS 2025.

• As an NLP Research Associate, I contributed to SemEval 2023 Task 3 by fine-tuning multilingual BERT models for news genre classification using PyTorch and Hugging Face, tackling challenges in multi-label and cross-lingual NLP.

• Previously, spent 3.5+ years as a Lead AI/ML Engineer at HighRadius Technologies, where I built and deployed ML pipelines for Fortune 500 clients like Nestlé, Uber, and Kraft Heinz. I implemented a GenAI-powered Smart Email Composer, improved deduction recovery by 12%, reduced DSO by 30%, and led a team of 10 across 15+ production ML projects and mentored over 50 interns in Data Science projects.

Technical Interests & Expertise

Large Language Models (LLMs)

• Fine-tuned and evaluated Transformer-based models like LLaMA, GPT, and BERT using LoRA, QLoRA, PEFT, and instruction tuning.
• Applied prompt engineering and task-specific adapters to optimize performance in constrained compute settings.

Generative AI & Agentic Systems

• Built RAG pipelines and agentic chatbots using LangChain, LlamaIndex, and OpenAI APIs.
• Integrated document-grounded reasoning and tool-calling to enable auditable, context-aware LLM interactions.

Machine Learning & Audio/Computer Vision

• Designed classification, regression, and sequence models for financial forecasting and medical applications.
• Engineered audio signal processing pipelines and developed knowledge distillation strategies for model compression and efficiency.

MLOps & Scalable Deployment

• Deployed models using Docker, Kubernetes, and Vertex AI across GCP, Azure, and AWS.
• Built CI/CD pipelines and reproducible workflows using MLflow, Weights & Biases, and cloud-native orchestration tools.

I enjoy building systems that make unstructured data usable, interpretable, and impactful.

Education

I hold a Master’s degree in Computer Science (Research Track) from the University of Colorado Boulder.

Actively Seeking

I am actively pursuing full-time opportunities in AI/ML, NLP, or GenAI roles. With experience in both research labs and enterprise environments, I’m excited to bring my expertise in production-grade AI systems.

📫 Feel free to reach out: sdwivedy30@gmail.com