Headline

Data & Applied Scientist-III at SAP Labs India || Tech & AI Content Creator || 150k+ Instagram @deepanshu_o7 || Mentor || Ex-TCS Digital || Silver Medalist..

About

I’m currently a Data Scientist at SAP Labs India, where I get to work on Embedded AI for SAP BTP (Business Technology Platform)—merging cutting-edge technology with real-world business impact. My journey into the world of AI began at TCS Research & Innovation as an ML Engineer, where I honed my skills in designing and deploying machine learning solutions. But my passion doesn’t stop at building AI models. I’m driven to help others find their path in tech. Through my Instagram page @deepanshu_o7, I share career strategies, actionable roadmaps, and tips for thriving in IT. It’s my way of giving back to a community that inspires me daily. I believe in balancing growth with curiosity, innovation with empathy, and work with purpose. Let’s connect if you’re passionate

Location

Bengaluru

AI stack

codex

opencode

Work experience

Senior Data & Applied Scientist at SAP, Mar 2026 - Present · 6 mos, Bengaluru, Karnataka, India · Hybrid, • Developed end-to-end GenAI-powered Anomaly Recommender using RAG, function calling, and CoT prompting, automating anomaly analysis and reducing resolution time by 75%. • Developed an agentic NL2SQL service using Multi-agent orchestration, RAG, and query execution in SAP HANA boosting analytics portal engagement by 20%

Data & Applied Scientist at SAP, Apr 2022 - Mar 2026 · 4 yrs, Bengaluru, Karnataka, India · Hybrid, • Architected multiseries time-series pipelines for 50+ tenant APIs enabling scalable anomaly detection and fore- casting in a multi-tenant environment. • Designed anomaly detection models using DBSCAN, Isolation Forest, IQR, LSTM, and Transformer autoencoders, with EVT thresholds and drift detection (PSI/MAE) for adaptive retraining. • Benchmarked forecasting models (ARIMA, LightGBM, N-HiTS, NeuralProphet) and solved cold-start forecasting using time-series representation learning and proxy similarity. • Led API behavior clustering using KMeans, Hierarchical, and DBSCAN, reducing anomaly false positives by 5% and improving forecast accuracy by 15%. • Built parallelized preprocessing pipelines and YAML-based ML experimentation framework for scalable training, backtesting, and hyperparameter tuning. • Managed CI/CD pipelines and secure production releases, enabling 8% customer upgrades to premium tiers via AI-driven capabilities.

Machine Learning Engineer at Tata Consultancy Services, Jun 2019 - Apr 2022 · 2 yrs 11 mos, India, 1. Performed Finetuning on the computer vision models and automated Tensorflow Object detection API 2.Achieved 90% reduction in data processing time by optimising Spark utilisation, reducing time for 1 TB from 28 days to 3 days. 3. Increased Model accuracy from 65% to 88% using feature engineering. 4. Developed Algorithm for Parallel Computing to optimise the model batch predictions reducing total time by 30%.

Education

Bachelor of Engineering, Mechanical Engineering, LNCT Group of Colleges

12th Standard, CBSE, Servite Convent Senior Secondary School