B
Bhanuprakash
Contact details — on requestProof of Work — on request
Experience
Z
Data Scientist
ZeoMinds IT Solutions · Mar 2024 – Dec 2025
- Developed an intelligent Legal Assistant using Generative AI with Retrieval-Augmented Generation (RAG), parsing and summarizing 358 legal sections from the Indian Penal Code via LangChain's WebBaseLoader, converting unstructured HTML into structured text files.
- Implemented RecursiveCharacterTextSplitter for document chunking (900-character segments with 180-character overlap), optimizing context retention and comprehension of legal materials.
- Generated high-accuracy vector embeddings using OpenAI's text-embedding-3-large model and built a Qdrant Vector Store for fast, scalable retrieval using hybrid search (semantic + keyword) with cosine similarity.
- Designed prompt pipelines integrating user queries, retrieved context, and system role messages, enabling GPT-40-mini to generate reliable, user-tailored legal outputs.
- Applied LLM Guardrails to validate model outputs, enforce compliance boundaries, filter hallucinations, and ensure safe, accurate responses preventing off-topic or legally unreliable answers from reaching end users.
- Integrated advanced retrieval techniques including Multi-Query Retrieval, SelfQuery, and Re-ranking to significantly enhance answer precision in legal Q&A tasks.
- Evaluated model quality using RAGAS metrics (Faithfulness, Relevance, Groundedness, Precision, Recall, MRR) to ensure high-quality legal responses.
- Applied Pydantic for structured response schema validation, enhancing type safety and reliability of the assistant's outputs.
- Built a multi-agent architecture with task-planning agents, enabling autonomous evidence retrieval, tool usage, and multi-step legal reasoning; achieved ~70% accuracy, reducing manual legal effort and turnaround time significantly.
Z
Jr. Data Scientist
ZeoMinds IT Solutions · Oct 2023 – Feb 2024
- Developed and deployed a GPT-3 / GPT-4 powered banking chatbot to automate customer queries (account balance, transaction history, loan eligibility, branch details), achieving 93.7% user satisfaction.
- Implemented NLP techniques for intent detection and entity recognition using transformer models (BERT), enabling accurate understanding of a wide range of customer inquiries.
- Deployed multi-turn dialogue management using reinforcement learning to ensure seamless conversations and personalized recommendations based on user history and preferences.
- Integrated secure token-based authentication and RESTful APIs to enable real-time data retrieval from core banking systems; deployed on AWS Lambda.
- Developed an intelligent RAG-based system to generate context-aware responses by retrieving relevant information from both structured and unstructured data sources.
- Achieved 93.7% user satisfaction by fine-tuning model performance and deploying continuous learning to adapt to customer feedback and new data.
N
Software Trainee
Ncore Technologies · Jun 2022 – Sep 2023
- Data Management & SQL: Managed data using SQL by creating tables, importing datasets, and building views and stored procedures to keep data organized and accessible.
- Extracted and analyzed sample populations with SQL queries (DML, DDL, JOIN, UNION) to support data-driven decision-making across the team.
- Data Analysis & Reporting: Developed weekly trend reports using Python libraries (Pandas, NumPy, Matplotlib, Seaborn) to transform raw data into actionable insights through analytics and visualization.
Skills 0 proven through work
Also works with
Confusion Matrix AnalysisTeamworkIssue ReportingProject CompletionTaking OwnershipEffective CommunicationAdaptabilityProject TrackingGoal UnderstandingWork Planning and PrioritizationTask AutomationModel Performance ImprovementLLM Cost OptimisationLLM-as-a-Judge EvaluationMulti-Agent System IntegrationData VisualizationSeaborn VisualizationMatplotlib VisualizationTabular Data Cleaning & ValidationMarket Trend AnalysisExploratory Data Analysis (EDA)NumPypandas UsagePython DevelopmentData ExtractionStored Procedure DevelopmentDatabase ViewsDatabase Table CreationData-Driven RecommendationsSQL ProgrammingRecommendation Feature DevelopmentCustomer Interaction AnalysisCustomer Behavior AnalysisAI-Powered PersonalizationReinforcement LearningUser FeedbackHyperparameter TuningAPI IntegrationConversational UX DesignUser Experience ImprovementGPT-3.5 Flash UsageApplication DeploymentSoftware Development PracticesRecall Metric InterpretationPrecision Metric InterpretationRagas Framework UsageAI Hallucination DetectionFalse Positive ReductionModel Metric Selection & InterpretationContent Moderation System DevelopmentDocument ChunkingUnstructured Data ProcessingUser Query AnalysisChatbot DevelopmentAI Model TestingAI Model ConfigurationCohere ReRanker UsageReciprocal Rank Fusion ImplementationQuadrant Vector Database UsageOpenAI UsageRetrieval-Augmented GenerationGenerative AI Application
Proof of Work
Proof of Work
Bhanuprakash shares this with people who ask. You'll hear back either way.
Education
Bachelor of Science (BSc), Statistics
A.V. College of Arts, Science and Commerce, Hyderabad · 2019 — 2022
Contact details
Contact details
Bhanuprakash shares this with people who ask. You'll hear back either way.
What drives their work
The Architect
Designs and constructs robust, reliable AI systems.
Bhanuprakash's superpower
You excel at engineering AI solutions that are both effective and trustworthy, ensuring they meet stringent requirements for accuracy and compliance.
How Bhanuprakash works
You would thrive in teams focused on developing, deploying, and maintaining robust, production-grade AI systems, particularly those requiring high accuracy and compliance.
Ask about this candidate
AI responses may contain errors. Proof status and source information are provided by Proof of Skill.
A Proof CV — one profile, kept current, with the proof attached. Back to the portal