Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Tejaswi Yadav is an AI/NLP Engineer based in Delhi, India, with 4+ years of experience building production-grade LLM applications, RAG systems, and multi-agent platforms. He has delivered AI solutions across recommendation systems, document automation, and enterprise workflow orchestration, with results including 98% recommendation accuracy and a 92% reduction in project grooming effort. His toolkit spans Python, LLM fine-tuning, agentic AI (LangGraph, MCP servers), vector databases, and computer vision. He specializes in turning ambiguous business problems into scalable, measurable AI systems.…Tejaswi Yadav is an AI/NLP Engineer based in Delhi, India, with 4+ years of experience building production-grade LLM applications, RAG systems, and multi-agent platforms. He has delivered AI solutions across recommendation systems, document automation, and enterprise workflow orchestration, with results including 98% recommendation accuracy and a 92% reduction in project grooming effort. His toolkit spans Python, LLM fine-tuning, agentic AI (LangGraph, MCP servers), vector databases, and computer vision. He specializes in turning ambiguous business problems into scalable, measurable AI systems.

Tejaswi Yadav

AI Engineer, Data Scientist, Software Engineer, +1





Tejaswi Yadav is an AI/NLP Engineer based in Delhi, India, with 4+ years of experience building production-grade LLM applications, RAG systems, and multi-agent platforms. He has delivered AI solutions across recommendation systems, document automation, and enterprise workflow orchestration, with results including 98% recommendation accuracy and a 92% reduction in project grooming effort. His toolkit spans Python, LLM fine-tuning, agentic AI (LangGraph, MCP servers), vector databases, and computer vision. He specializes in turning ambiguous business problems into scalable, measurable AI systems.…Tejaswi Yadav is an AI/NLP Engineer based in Delhi, India, with 4+ years of experience building production-grade LLM applications, RAG systems, and multi-agent platforms. He has delivered AI solutions across recommendation systems, document automation, and enterprise workflow orchestration, with results including 98% recommendation accuracy and a 92% reduction in project grooming effort. His toolkit spans Python, LLM fine-tuning, agentic AI (LangGraph, MCP servers), vector databases, and computer vision. He specializes in turning ambiguous business problems into scalable, measurable AI systems.

Available to hire

Tejaswi Yadav is an AI/NLP Engineer based in Delhi, India, with 4+ years of experience building production-grade LLM applications, RAG systems, and multi-agent platforms. He has delivered AI solutions across recommendation systems, document automation, and enterprise workflow orchestration, with results including 98% recommendation accuracy and a 92% reduction in project grooming effort. His toolkit spans Python, LLM fine-tuning, agentic AI (LangGraph, MCP servers), vector databases, and computer vision. He specializes in turning ambiguous business problems into scalable, measurable AI systems.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Work Experience

Software Developer (AI) at Tru0 4

April 10, 2025 - Present

Developed an AI-powered meal planning and recipe recommendation system using Milvus Vector DB, BGE search, Voyager re-ranker, and Gemini model; implemented price-per-portion calculations and intelligent recipe ranking achieving 98% recommendation accuracy and improving business decision-making. Built a Multi-Agent AI platform using BRD/LOE and architecture agents to automate BRD/user story generation, effort estimation using historical data, and high-level architecture creation, reducing project grooming effort by 92%. Developed an AI-powered checklist utility to extract requirements and objectives from technical standard PDFs using Mistral for structured data/table extraction and integrated Azure Foundry GPT to standardize and refine outputs (over 95% extraction accuracy across complex formats). Built a Quote History tool leveraging Azure Cognitive Search DB with semantic similarity search for intelligent quote recommendations, improving contextual retrieval accuracy by 90% and decisi

Software Developer (AI) at Tru04

April 1, 2025 - April 1, 2025

Developed an AI-powered meal planning & recipe recommendation system using Milvus Vector DB, BGE search, Voyager reranker and Gemini model to recommend recipes based on ingredients, business rules, and user requirements. Implemented price-per-portion calculations and intelligent recipe ranking achieving 98% recommendation accuracy while improving business decision-making. Built a multi-agent AI platform with BRD/LOE generation and architecture agents to automate BRD/user story generation, effort estimation using historical data, and high-level architecture creation—reducing project grooming effort by over 92% and accelerating vendor assignment and project planning. Developed an AI-powered checklist utility to extract requirements and objectives from diverse technical standard PDFs. Used Mistral AI OCR for structured data/table extraction and integrated Azure Foundry GPT for output standardization and refinement, achieving over 95% extraction accuracy across complex document formats.

AI/NLP Engineer at Taxman NTechnologies Private Limited

July 1, 2024 - July 1, 2024

Built domain-specific LLM applications using continued pretraining and LoRA-based fine-tuning to improve performance while minimizing training costs. Developed a Retrieval-Augmented Generation pipeline using parent-child retrieval, hybrid search (sparse + dense embeddings) and Voyager reranker, improving accuracy and response quality with Llama 3.28B. Created scalable book recommendation systems using Python, TF Vectorizer and cosine similarity (scikit-learn), with a PostgreSQL backend for optimized data communication. Automated content creation processes by implementing OCR, Pydantic and LLM models to fully eliminate manual intervention for content writing. Implemented global search for subscribed users enabling retrieval of top documents matching search criteria using RAG and hybrid search. Developed Graph RAG architecture leveraging knowledge graph relationships for enhanced document retrieval—improving answer relevance and context preservation by 40% using LangChain, LangChain-op

AI/NLP Engineer at TA XM AN N T E CH NOLO G IE S P R IV ATE L IM IT E D

July 1, 2024 - Present

Developed domain-specific LLMs by continued pretraining and LoRA-based fine-tuning to improve performance while minimizing training costs. Created a custom Retrieval-Augmented Generation pipeline using parent-child retrieval, hybrid search (sparse + dense embeddings), and Voyage Reranker; enhanced accuracy and response quality using Llama 3.28B. Implemented scalable book recommendation system using Python, TF Vectorizer, and cosine similarity from scikit-learn with PostgreSQL for optimized database communication. Automated content creation using OCR, Pydantic, and LLM models (e.g., GPT) to eliminate manual intervention. Built global search feature for subscribed users using RAG and hybrid search to retrieve top matching documents. Implemented Graph RAG architecture leveraging knowledge graph relationships for improved document retrieval and answer relevance/context preservation using LangChain/LangChain-OpenAI/LLM Graph Transformer/Neo4j/Graph Database (about 40% improvement). Develope

Software Analyst at OnGraph Technologies Limited

September 1, 2023 - Present

Developed an invoice management system using Django and Python, integrating OCR (DocTr) and Cloud-3 Sonnet AI for automated data extraction, improving accuracy by 40%. Built a heart disease prediction system using Python, Pandas, Matplotlib/Seaborn, scikit-learn with 6 ML models (logistic regression, decision trees, random forest, neural networks, SVM, and others) and integrated GPT Turbo to generate user-friendly risk explanations. Developed real-time car damage detection using YOLOv8, OpenCV, and WebRTC, improving model accuracy from 60% to 80% via data augmentation and fine-tuning. Built an AI interview chatbot using LangChain and OpenAI integrated with a vector database for personalized feedback and conversation tracking to help students and job seekers improve interview skills.

Data Analyst at Savvy Enterprises

January 1, 2021 - September 1, 2023

Managed and cleaned datasets from multiple sources to ensure data accuracy and consistency. Performed SQL querying to extract insights from relational databases, optimized query performance and handled data pipelines. Conducted exploratory data analysis and generated actionable insights to support business decisions. Built visualizations and dashboards using Tableau to track key business metrics. Provided data-driven recommendations across marketing, sales and finance to improve business performance.

Data Analyst at Savvy Enterprise

January 1, 2021 - Present

Managed and cleaned datasets from multiple sources to ensure accuracy and consistency. Wrote SQL queries to extract insights from relational databases, optimize query performance, and handle data pipelines. Performed exploratory data analysis and generated actionable insights to support business decisions. Created dashboards and visualizations using Tableau to track key business metrics. Provided data-driven recommendations across marketing, sales, and finance for improving business performance.