hi, Abhisek here.
Data engineering · Germany
I design cloud data platforms, automated pipelines, analytics-ready models, and decision tools—from enterprise-scale systems to connected-vehicle telemetry.
Three systems that show how I structure messy, high-volume signals into reliable data products.
A production-oriented lakehouse for connected-vehicle telemetry, designed to support batch and streaming ingestion, data quality, governance, analytics, and future ML workloads.
An end-to-end data product that converts battery and charging events into trustworthy datasets for health analysis, charging behavior, warranty risk, and fleet operations.
ProblemVehicle events arrive with different quality, timing, and operational meaning.
ApproachMedallion layers create traceable transformations from raw events to analytics-ready products.
Next layerTime-series features and an early-warning model for battery degradation.
An AI-assisted report-assessment solution developed in ServiceNow for the Norwegian Radiation and Nuclear Safety Authority during my NTNU exchange.
ContextA consequential domain where AI must support—not obscure—expert judgment.
SignalExperience translating an institutional workflow into an AI-supported solution.
PrincipleKeep traceability, human review, and clear limitations visible in the workflow.
Professional data engineering, systems research, and a new role in automotive digitalization beginning October 2026.
Bosch · Automotive Procurement Digitalization
Selected to help build structured data models, automated pipelines, analytics, and trusted data foundations for AI and advanced analytics.
Mphasis Ltd
Designed and operated enterprise data pipelines across cloud, analytics, and reporting environments.
Building toward production AI
My foundation is the infrastructure beneath intelligent systems: ingestion, transformation, quality, orchestration, and monitoring. I’m extending that foundation into retrieval, evaluation, and ML deployment.
A grounded assistant connecting vehicle telemetry with maintenance knowledge through retrieval, citations, and tool-driven queries.
A time-series ML layer for the EV platform, with leakage-aware validation, operational metrics, drift monitoring, and a model card.
Status labels are intentional: future work is never presented as completed experience.
Technische Universität Ilmenau
Software safety · systems engineering · advanced databases · algorithmsNorwegian University of Science & Technology
Software architecture · recommender systems · programming & numericsIIIT Delhi
Statistics · modeling · machine learning · data visualizationOpen channel · Germany
I’m interested in data, platform, and applied-AI engineering work where trustworthy infrastructure matters.