Headline

AI/ML Engineer @ SenseOps | Data Science & ML | NLP & LLMs | Python • SQL • Power BI | Built 95% Accurate NLP-to-SQL System | Open to DS & AI Roles

About

AI/ML Engineer building production-ready NLP and generative AI systems that transform business processes through intelligent automation and data-driven decision-making. Core Expertise: I specialize in large language models (LLMs), natural language processing (NLP), retrieval-augmented generation (RAG), machine learning model development, and data pipeline engineering. Proficient in Python, SQL, BERT, Prompt Engineering, FastAPI, and cloud deployment. Currently architecting AI solutions at SenseOps focused on EV battery analytics and real-time monitoring systems. Key Achievements: Engineered 95% accurate NLP-to-SQL system using LLaMA and Rasa, enabling non-technical users to query complex databases in natural language—reducing reporting time by 50% Designed self-learning retrieval framework using ChromaDB, decreasing query synthesis latency by 80% for high-frequency business requests through intelligent caching Built real-time health monitoring engine for 1,000+ EV battery units using FastAPI and advanced SQL window functions, accelerating maintenance response times by 50% Developed offline speech-to-speech translator for military training across 4 languages (Hindi, Telugu, Tamil, Bengali), achieving <10% WER and >85% BLEU translation quality; reduced interpreter costs by >50% annually Created anomaly detection system analyzing 616K+ time-series records, achieving >90% detection accuracy and preventing ~25% of unplanned infrastructure downtime Built BERT-based recruitment intelligence system processing 8,000 interview transcripts, reducing candidate screening time by 50% and evaluation costs by ~40% Technical Skills: Python • SQL • BERT • LLaMA • Rasa • FastAPI • ChromaDB • Prompt Engineering • Machine Learning • Deep Learning • Data Engineering • ETL Pipeline Development • Power BI • Git • Docker Current Focus: Seeking full-time data science or AI/ML engineer opportunities in Bengaluru where I can design and deploy production-grade AI systems that directly impact business metrics and user experience. Passionate about building scalable ML infrastructure, optimizing model performance, and translating complex AI concepts into actionable business solutions. Open to: Full-time roles in data science, AI/ML engineering, LLM engineering, and related positions.

Location

India

AI stack

hugging face

langchain

langgraph

ollama

Work experience

AI/ML engineer at SENSEOPS TECH SOLUTIONS PVT LTD, Jul 2025 - Present · 1 yr, Bengaluru, • Designed and implemented a 95% accurate NLP-to-SQL system using LLaMA and Rasa to enable natural language querying over EV battery databases. • Integrated LLM pipelines with structured SQL databases for real-time query execution. • Applied prompt engineering and response validation techniques to improve model reliability. • Assisted in backend query optimization to enhance system performance. • Collaborated with cross-functional teams to translate business requirements into AI-driven solutions. Technologies: Python, SQL (SSMS), LLaMA, Rasa, NLP, Large Language Models (LLMs), Prompt Engineering, REST APIs

Machine Learning Engineer at SENSEOPS TECH SOLUTIONS PVT LTD, Jul 2025 - Present · 1 yr, • Architected an agentic NLP-to-SQL chatbot for non-technical manufacturing workers that cut reporting time by 50% across 50K+ daily records. Key challenge was LLM hallucinating column names; resolved by injecting live schema context with multi-shot prompting, achieving 95% query accuracy • Implemented a RAG evaluation framework using RAGAS and LLM-as-judge (faithfulness, answer relevancy, context precision) to set production-grade benchmarks. When retrieval returned irrelevant domain-specific results, redesigned the pipeline with hybrid BM25 and semantic search plus result re-ranking, eliminating irrelevant matches • Deployed a real-time EV battery manufacturing dashboard (FastAPI + web frontend) for floor workers surfacing shift, day, and week insights with automated failure alerts and root cause reasons across 1,000+ battery units. Replaced an unusable CLI tool by building a worker-friendly UI with auto-refresh and zone-based alerting • Engineered a prompt engineering pipeline with multi-shot learning, MLflow experiment tracking, and response validation, boosting query accuracy from 85% to 95% and cutting false positives by 75% • Optimized SQL query performance via indexing and query plan analysis, lifting throughput by 35% and reducing CPU utilization by 28%

Data Science Intern at 360DigiTMG, Jan 2025 - Jun 2025 · 6 mos, India, • Diagnosed solar power anomalies for an alloy manufacturer by analyzing 175,640 inverter records and 89,784 weather observations across 4 data sources using EDA, datetime merging, log transformation, and threshold-based filtering. • Engineered features using RobustScaler and PCA, then trained Isolation Forest, One-Class SVM, and Local Outlier Factor models, achieving 90%+ detection accuracy, 25% anomaly reduction, and 40% infrastructure cost savings. • Built an automated interview transcript analysis system on 8,000 recruitment records using text preprocessing, TF-IDF vectorization, SMOTE, and fine-tuned BERT, achieving 87% classification accuracy. • Developed an extractive summarization pipeline using attention mechanisms to generate concise candidate summaries, reducing screening time by 55% and evaluation costs by 40%. • Deployed both solutions as Streamlit dashboards for real-time solar monitoring and recruitment insights, enabling faster data-driven decisions for operations and HR teams. Technologies: Python, Machine Learning, Deep Learning, NLP, BERT, Hugging Face, TF-IDF, SMOTE, Isolation Forest, One-Class SVM, LOF, PCA, RobustScaler, Scikit-learn, Pandas, NumPy, Streamlit, NLTK, Matplotlib, Seaborn, Pickle

Data Science Intern at NullClass, May 2024 - Aug 2024 · 4 mos, Remote, • Developed a real-time multilingual translation web application using Flask and Transformer-based encoder-decoder models, supporting 4 languages with automatic language detection and dynamic routing. • Built a cross-lingual semantic search and multilingual sentiment analysis system using sentence-transformers, vector embeddings, and mBERT for language-independent retrieval and classification. • Trained and optimized sequence-to-sequence NLP models using BLEU score, accuracy metrics, and hyperparameter tuning to improve translation quality. • Automated end-to-end NLP pipelines including preprocessing, feature engineering, and deployment of trained models into interactive web interfaces using Python, Pandas, and Scikit-learn. Technologies: Python, Machine Learning, Natural Language Processing (NLP), Transformers, mBERT, Sentence Transformers, Flask, Pandas, Scikit-learn, Deep Learning, Vector Embeddings, Semantic Search, Hyperparameter Tuning

Education

Bachelor of Technology, Computer Engineering, PES University