Abhinav Sharma

Abhinav Sharma PROFESSIONAL SUMMARY Product building with robust experience across Data Analytics ,ML and Generative AI. Experienced in identifying emerging market opportunities, validating technical feasibility, and developing digital products from concept to prototype. EDUCATION B.Tech 2015 IIT BHU M.Tech 2017 IIT Kharagpur TECHNICAL SKILLS Programming Languages: Python, SQL GenAI: LangChain, LangGraph, Agentic Data Processing: Pandas, NumPy Visualization & Reporting: Matplotlib, Power BI, MS Excel Cloud Platforms: Azure ML MLOps Tools: MLflow, Git Databases: MySQL, Snowflake ML Techniques: Supervised/Unsupervised Learning, Deep Learning, Reinforcement Learning Statistics: Hypothesis Testing, A/B Testing, Bayesian Methods, Experimental Design CLOUD TECHNOLOGIES Azure PROFESSIONAL EXPERIENCE Capgemini – Consultant Duration: July 2021– Present Location: Hybrid As part of Capgemini’s advanced analytics and AI team, I played a pivotal role in designing and deploying end-to-end data-driven solutions across sectors such as banking, healthcare, retail, and insurance. My work bridged the gap between business strategy and technical implementation, particularly in the domains of maths & analytics, machine learning, Generative AI, and Azure cloud services. Key Responsibilities & Achievements: End-to-End Machine Learning & Statistical Modeling o Designed and implemented data-driven ML pipelines for fraud detection, churn prediction, and demand forecasting using models such as Logistic Regression, Random Forests, XGBoost, and LSTMs. o Built automated ML workflows with MLflow for experiment tracking, model versioning, and performance reproducibility. o o o o o o o o Performed statistical inference and uncertainty quantification using confidence intervals, bootstrap sampling, and hypothesis testing to assess predictive model performance. Developed a personalized recommendation engine using collaborative filtering, matrix factorization, and deep autoencoders, boosting user engagement by 30%. Led a cross-functional ML team to deliver a scalable predictive analytics platform, integrating data preprocessing, model training, and real-time inference. Conducted A/B testing, statistical hypothesis testing, and causal impact analysis to validate model uplift and quantify business ROI. Enhanced model accuracy via feature selection (LASSO, PCA) and Bayesian hyperparameter optimization, reducing overfitting and improving generalization. Deployed production models on Azure ML Studio with CI/CD pipelines, implementing MLOps practices for monitoring and retraining. Integrated LLM-based modules (e.g., GPT-family, BERT embeddings) for unstructured text processing, entity extraction, and contextual insights from claims and customer feedback. Partnered with engineering teams to build real-time inference APIs and feedback loops for continuous model improvement. LLM & GenAI Projects o Built and deployed Generative AI solutions (chatbots, summarizers, auto-response agents) using Azure OpenAI. o Designed prompt engineering and RAG pipelines to improve accuracy, tone, and contextual relevance. o Integrated GenAI services via Azure Functions and Web Apps for secure, scalable deployment. o Conversational AI o Developed hybrid rule-based + LLM chatbots for HR, customer support, and finance workflows. o Enhanced conversational context using vector search, APIs, and memory components. AI-Powered Product Search Assistant o Built an LLM-driven product search engine that converts natural language queries into structured parameters for precise retrieval. o Implemented semantic parsing pipelines to extract and map attributes (category, brand, price, features) to catalog fields with synonym and multilingual handling. o Integrated hybrid search (keyword + vector) using schema-aware embeddings to improve precision and recall across millions of products. o Added LLM-based reasoning and guardrails for query disambiguation and clarification prompts. o Engineered a feedback loop to refine prompts and boost parameter extraction accuracy over time. o Delivered real-time, high-precision search, reducing manual filtering by 60%, with enterprise scalability, RBAC, and secure API integration. Text-to-SQL Agentic AI System o Built an LLM-powered agent that converts natural language queries into optimized SQL across MySQL, PostgreSQL, and Snowflake. o Engineered an agentic RAG workflow for schema parsing, query validation, and self-correcting reasoning to ensure accuracy and robustness. o Designed adaptive schema-parsing pipelines handling large, complex database structures with contextual awareness. o Implemented guardrails to prevent hallucinations, manage ambiguity, and auto-reprompt users. o Optimized SQL generation via query planning, cost estimation, and caching for performance gains. o Used LangChain, Python, and Azure OpenAI APIs to orchestrate multi-step reasoning agents with monitoring dashboards for accuracy and latency. o Fine-tuned and prompt-engineered models on domain data, achieving >90% query accuracy and reducing manual SQL workload by 70%. o Deployed as a scalable, multi-tenant enterprise system with RBAC and audit logging. Data Engineering & Analytics o Led the structured and unstructured data transformation process using SQL, and Python to prepare high-quality datasets for analytics. o Created dashboards and visual reports using Power BI to communicate insights to business stakeholders. Client & Team Leadership o Served as technical lead on multiple projects, managing junior data scientists and aligning delivery with client expectations. o Worked closely with product owners, domain experts, and tech architects to shape solutions from ideation to deployment. Snapmint Credit Advisory – Analyst Duration: Dec 2019 – May 2020 Location: Mumbai At Snapmint — a fintech company providing zero-cost EMIs on consumer electronics — I contributed to building a data-centric credit evaluation system that transformed how consumers accessed credit without traditional documentation. My role centered around leveraging machine learning, consumer behavior data, and risk analytics to drive smarter, faster, and more inclusive lending decisions. Key Responsibilities & Achievements: Credit Scoring Model Development o Built and deployed machine learning models to predict customer creditworthiness. o Used transactional behaviour, device metadata, alternative data (e.g., social signals), and credit bureau inputs to engineer predictive features. o Achieved a 22% lift in precision for risk classification, reducing default rates significantly without cutting approval volumes. Ad Campaign Analytics & ROI Optimization o Designed and implemented data-driven ad performance models to measure ROI across digital channels (Google Ads, Meta, and programmatic platforms). o Built predictive models using regression and uplift modelling to estimate conversion lift and customer lifetime value (CLV) from campaign exposure. o Conducted A/B testing and multivariate experiments to identify high-performing creatives, audience segments, and bidding strategies. o Developed automated dashboards (Power BI / Tableau) for real-time tracking of KPIs — CTR, CPC, CPA, ROAS, and engagement metrics. o Applied attribution modelling (Markov Chains, time decay) to quantify channel contribution to conversions and optimize media spend allocation. o Leveraged Python (pandas, scikit-learn, statsmodels) for campaign data cleaning, feature engineering, and performance forecasting. o Created marketing mix models (MMM) to analyse offline and online campaign synergy and guide budget optimization decisions. o Collaborated with marketing and product teams to reduce cost-per-acquisition by 25% and increase campaign ROI by 40% through advanced targeting and spend optimization. o Built customer segmentation and clustering models using K-means to personalize ad messaging and improve engagement rates. o Presented analytical insights and actionable recommendations to leadership for data-driven marketing strategy formulation. Risk Segmentation & Personalization o Created customer segmentation models to dynamically assign credit limits and EMI options. o Enabled tiered risk-based pricing by clustering customers based on risk propensity and repayment behavior. Real-Time Analytics & Deployment o Developed lightweight scoring APIs for real-time credit decisions at point-of-sale (online and offline), ensuring sub-second latency. o Collaborated with engineering teams to deploy models using Flask APIs on cloud infrastructure. Business Intelligence & Strategy Support o Designed executive dashboards in Power BI for tracking approval trends, repayment patterns, and delinquency rates. o Provided data-driven insights that influenced product decisions, such as eligibility criteria refinement and new EMI slab creation. Fraud Detection Enhancements o Identified fraudulent patterns using anomaly detection and rule-based filters, improving risk flagging accuracy. o Reduced early-stage fraud losses by ~18% within one quarter. PYE ELECTRO SYSTEMS - Consultant Duration: Jan 2019 – Dec 2019 Location: Bhopal Applied data-driven analysis to support the planning and execution of residential and commercial solar energy projects across Madhya Pradesh. Leveraged data insights to optimize system design, improve financial feasibility, and enhance customer acquisition strategies. Key Contributions: Feasibility & Performance Analytics o Conducted quantitative assessments of site potential using solar irradiation, consumption, and cost data to evaluate ROI and payback periods. o Built analytical models to simulate generation performance and forecast long-term savings, enabling 50–60% reduction in client energy costs. o Used Excel/Python-based analysis for system sizing, yield estimation, and tariff impact modeling. Customer & Market Insights o Analyzed customer profiles and property attributes to identify high-potential solar adoption segments, supporting targeted outreach strategies. o Designed dashboards to track project pipeline, lead conversion, and energy generation metrics, improving visibility into business performance. Operational Optimization o Collaborated with the operations team to analyze installation timelines and workflow data, reducing average project delivery time. o Applied data insights to resource allocation and vendor performance evaluation, enhancing reliability and service quality. ABM (Association of Billion Minds) – Political Data Analyst Duration: Nov 2018 – Dec 2018 Location: Delhi At ABM — a leading political consultancy firm — I served as a core data strategist, advising India’s top political parties through in-depth analytics, demographic modeling, and historical voting behaviour analysis. My work helped shape constituency-level strategies, messaging, and campaign decisions backed by millions of data points across elections, geographies, and communities. Key Responsibilities & Achievements: Voter & Demographic Analysis o Analyzed voter rolls, census data, and historical election results to identify vote swing patterns, core voter segments, and micro-targeting opportunities. o Created demographic heatmaps and segmentation models at the booth and constituency level to support precision targeting. Election Strategy & War Room Support o Provided data-driven insights to political war rooms to optimize on-ground mobilization, resource allocation, and messaging. o Contributed to real-time dashboards during campaign cycles to monitor ground feedback, turnout trends, and sentiment shifts. Behavioural & Sentiment Insights o Used NLP techniques to analyze social media data, local news coverage, and survey transcripts for constituent sentiment tracking. o Delivered periodic reports on local issues, leader perception, and public engagement, enabling proactive campaign adjustments. Survey & Field Intelligence Integration o Designed data models to ingest and synthesize inputs from field surveys, door-to-door campaigns, and telephonic outreach. o Developed triangulation logic to validate field reports against historical data and predictive models. Predictive Modeling for Election Outcomes o Built logistic regression and ensemble models to forecast win probabilities based on candidate performance, demographic trends, and turnout history. o Flagged high-risk constituencies early in the campaign, allowing strategic intervention. FutureBridge – Market Intelligence Analyst Duration: Sept 2017 – Oct 2018 Location: Mumbai At FutureBridge, I worked as a Market Intelligence Analyst, delivering strategic insights to Fortune 500 clients across industries such as automotive, energy, chemicals, healthcare, and consumer goods. My role involved conducting deep quantitative and qualitative research to uncover innovation trends, competitive landscapes, and market entry opportunities. Key Responsibilities & Achievements: Market Opportunity Identification o Conducted in-depth market sizing, growth forecasting, and competitive benchmarking for clients looking to enter new geographies or launch new products. o Identified emerging trends and disruptions using patent analysis, funding data, and technology adoption curves. Technology & Innovation Scouting o Built innovation intelligence reports by analyzing R&D pipelines, academic research, and startup ecosystems. o Evaluated TRLs (Technology Readiness Levels) and innovation maturity to advise clients on partnership or investment decisions. Voice of Customer & Expert Interviews o Coordinated and synthesized findings from primary research—including interviews with industry experts, stakeholders, and end-users. o Merged insights from customer feedback and field intelligence to validate strategic assumptions. Data Synthesis & Analytics o Used Excel, SQL, and Python for cleaning, structuring, and analyzing large-scale industry datasets. o Designed compelling storyboards and data visualizations to deliver insights that influence board-level decisions. Client-Facing Consulting Reports o Delivered executive reports including go-to-market strategies, SWOT analyses, and competitive positioning maps. o Frequently collaborated with global consulting teams to tailor findings to client-specific business models and regions.