Kapil Jogi

Data Analyst and Machine Learning Engineer with 5 years of experience building production ML and Agentic AI systems - from LLM-powered autonomous analytics (Gemini, LangChain, LangGraph, RAG) to time series forecasting and classical ML. Skilled at designing end-to-end pipelines from data ingestion through model deployment, and at turning ambiguous business problems into automated, measurable analytical systems. Experienced owning projects from raw data to a working prompt-to-SQL / voice-driven insight product.

Data Analyst - ETP International Pvt. Ltd. - Mumbai

(2024-07)

• Owned full-lifecycle analytical projects end-to-end - ingestion, preprocessing, model development, evaluation, and deployment - delivering 3 production models/reports used by Business Analysts and Customers.
• Built regression, classification, and clustering models with cross-validation and hyperparameter tuning, improving precision, recall, and F1 score over the prior baseline.
• Developed a Facebook Prophet-based sales forecasting pipeline incorporating seasonality, holiday effects, lag features, and anomaly detection, reducing forecast error (MAPE) by 18%.
• Designed and shipped an Agentic AI system (Gemini LLM + LangChain + RAG) that autonomously retrieves data, reasons over it, and generates business insights and reports - cutting manual reporting time by 30%; built an evaluation set to track answer accuracy/hallucination rate and iterated on retrieval quality and prompt design.
• Extended the platform with a voice-enabled interface (SpeechRecognition, PyAudio, gTTS) and a prompt-to-SQL pipeline, converting natural-language speech into validated SQL, executing against the production schema, and returning AI-generated explanations with guardrails against unsafe/incorrect query generation.
• Built and maintained end-to-end ETL pipelines (cleaning, feature engineering, outlier treatment, normalization, encoding) processing 10 lakhs+ records into production-ready datasets.
• Partnered with product, engineering, and business stakeholders to translate analytical output into scalable deployments, presenting tradeoffs and results to non-technical audiences.

Machine Learning Engineer - Mobicule Technologies Pvt. Ltd. - Mumbai

(2021-07 - 2024-06)

• Built machine learning models for telecom customer churn prediction using demographic, account, and service-usage features, improving churn identification by [26%] and directly informing retention campaigns.
• Analyzed large-scale telecom datasets to surface churn drivers, delivering insights that influenced [Repeat Purchase Rate].
• Wrote production-quality ML code following engineering best practices (version control, code review, testing), contributing reusable components adopted across [3] projects.
• Conducted data quality assessments and statistical analysis, presenting findings directly to project leads and stakeholders to guide roadmap decisions.
• Built OCR workflows (Tesseract + OpenCV) for text extraction and image preprocessing from scanned documents, improving extraction accuracy to [89%] and reducing manual data-entry time by [15%].
• Authored technical documentation for system architecture, algorithms, and backend components, reducing onboarding time for new engineers on the OCR/analytics codebase.

MSc - Computer Science - BAMU Aurangabad (2021)

BCA - Management Science - BAMU Aurangabad (2019)