Sahil Bharti

RAG Data Scientist

Cohere

Jun 2024 - Mar 2025 (10 months)

• Integrated RAG into the Command model to boost LLM performance on code generation tasks, improving SQL/Python completion accuracy through targeted dataset curation and fine-tuning. • Built and managed scalable annotation pipelines for code tasks (semantic labeling, query disambiguation, multi-hop retrieval), improving model robustness in production developer workflows. • Led a team of 6 annotators, implemented sprint-based workflows and KPIs (data quality, throughput, efficiency), driving a 30% productivity increase in two quarters. • Developed evaluation frameworks combining automated and human-in-the-loop metrics to track LLM gains in retrieval accuracy, code correctness, and NL interpretability. • Used topic modeling (LDA) on inte

D

Data Scientist (Volunteer)

Digna AI

Apr 2024 - Feb 2025 (11 months)

• Directed the integration of OpenAI's API with RAG to enhance a chatbot assisting war affected Ukrainians with relocation to Norway, boosting user interaction accuracy. • .Employed Python and Git within an Agile framework to manage code effectively, ensuring rapid iteration and high-quality updates in a collaborative, volunteer-based project environment. • Managed the migration of the project to Amazon Bedrock to leverage its superior LLM capabilities, aiming to enhance AI performance and scalability.

P

PDF Assistant Chatbot

Dec 2023 - Jan 2024 (2 months)

• Leveraged Llamaindex, RAG, and OpenAI API to develop a comprehensive document analysis and question-answering system. • Successfully interacted with 4+ PDFs, extracting relevant data, and enabling users to interact with the PDF in real-time.

E

Exploring Job Description-Skills Relationship

Mar 2023 - May 2023 (3 months)

• Conducted comprehensive analysis of job-technology skill relationships within a scalable Big Data framework, merging diverse sources for seamless updates. • Employed advanced data processing in Databricks, using DataFrames, RDDs, and SQL to visualize connections across 500+ job roles and skills for accurate dataset alignment. • Presented findings from analysis of over 50,000 Job Descriptions and O*NET datasets, revealing key trends and correlations, demonstrating the project's relevance to real-world job market insights.

Data Scientist

Navisite

Jun 2020 - Sep 2022 (2 years 4 months)

• Implemented POC AI-driven chatbot for the website, leveraging LLMs (GPT 3) to enhance customer engagement and support. • Developed and deployed a text classification system for customer support ticket prioritization, utilizing scikit-learn, NLTK, and Azure achieving a 20% reduction in average response times. • .Utilized Azure DevOps for sentiment analysis model deployment, achieving 80% faster integration and testing via CI/CD pipelines, while ensuring streamlined project workflows with Git for efficient version control, enhancing service quality. • Employed classification algorithms to analyze extensive customer behaviour trends, accurately predicting churn risk within the company's diverse client base reducing churn by 20%.

Data Analyst

Navisite

Nov 2019 - Jun 2020 (8 months)

• Identified inefficient queries and optimized them using indexing techniques, reducing query response time by 15% and improving data accessibility for downstream analytics. • Analyzed SQL logs and database metrics to diagnose client issues, and implemented solutions like data partitioning or schema changes, resulting in a 100% resolution rate and improved customer satisfaction. • Automated database backups, replication, and failover for over 20 stakeholders, minimizing downtime to less than 1% and safeguarding critical data integrity.

About

Detail

Timeline

Résumé