Available for opportunities

Hi, I'm Vishwanath.
I build scalable backends & distributed systems.

Masters Student in Computer Science at SRH Berlin. I specialize in designing APIs, managing data pipelines, and deploying to the cloud. Passionate about AI/ML integration and maintainable code.

Vishwanath A Sardeshpande

Experience

Hitachi Rail

ML Engineer — Working Student

Hitachi Rail · Berlin, Germany

Sep 2025 – Feb 2026
  • Conducted a comparative analysis of state-of-the-art GNN models (GAT & GCN), benchmarking on synthetic datasets via NetworkX to identify the optimal architecture for graph generation.
  • Engineered custom GNN architectures in PyTorch, including generative models trained from scratch to produce spanning trees with 95% accuracy, outperforming baseline GCN models.
  • Developed a custom loss function and reward system for supervised GNN training, enhancing model convergence and enabling new predictive capabilities on large-scale interconnected data.
PyTorch GNNs NetworkX PyTorch Geometric
Wayfair

Backend Developer — Working Student

Wayfair · Berlin, Germany

Aug 2024 – Apr 2025
  • Designed and implemented Vertex AI Pipelines on GCP to automate ML workflows, including ad-hoc image annotation using Generative AI models to build and label datasets.
  • Managed job scheduling and debugging of legacy DAGs in Apache Airflow during migration to Vertex AI, automating complex workflows to reduce errors and support seamless transitions.
  • Built APM and system health dashboards in Datadog and Looker Studio, tracking P99/P95/P75 latencies and dataset generation from BigQuery dumps for real-time operational insights.
GCP Vertex AI Apache Airflow Python Datadog BigQuery
Kennametal

SAP ABAP Developer

Kennametal Shared Service Pvt. Ltd. · Bengaluru, India

May 2022 – Mar 2024
  • Designed and developed Smart Forms, Adobe Forms, and custom reports using ABAP to meet specific user requirements and support business operations.
  • Leveraged AI and LLM technologies to build a warehouse management chatbot, improving operational efficiency by 18% through streamlined workflows.
  • Developed an OData application enabling seamless cross-platform integration and efficient data exchange.
SAP ABAP AI/LLM OData Smart Forms

Featured Projects

Master's Thesis

Self-Reflective Autonomous Driving

Designed a teacher–student framework where a rule-based policy generates supervision for a VLM student. Trained with SFT and reinforcement learning in CARLA using GRPO, with a self-reflection mechanism for failure diagnosis.

Vision-Language Models CARLA GRPO PyTorch

Perception–Reasoning–Action Pipeline

GenAI Vision Assistant

Built a LangGraph vision agent with Google Gemini VLM for complex scene analysis and multi-step visual reasoning. Deployed with Docker and integrated Kokoro TTS for low-latency audio output to a mobile client.

LangGraph FastAPI Gemini VLM Docker

React + FastAPI

Full-Stack E-Commerce

E-commerce platform with a React frontend, FastAPI backend, and PostgreSQL with transactional order handling. JWT-based auth for customers and admins, fully dockerized for production.

React FastAPI PostgreSQL Docker

Technical Skills

Languages

  • Python
  • JavaScript / TypeScript
  • SQL
  • SAP ABAP

Backend

  • FastAPI
  • Django
  • Flask
  • Microservices

AI / Data

  • PyTorch / TensorFlow
  • LangChain / LangGraph
  • PostgreSQL / MongoDB
  • Neo4j / ChromaDB

Cloud & Ops

  • Azure & GCP
  • Docker & Kubernetes
  • GitHub Actions
  • Vertex AI

Education

M.S. Computer Science — Big Data & AI

SRH Berlin University of Applied Science · Berlin, Germany

Apr 2024 – Mar 2026

B.E. Information Science & Engineering

Sri Krishna Institute of Technology · Bangalore, India

Aug 2022