Altamash Khan

Altamash Khan

Did the war forge the spear that remained? No. All it did was identify the spear that wouldn't break

Altamash Khan

Altamash Khan

Did the war forge the spear that remained? No. All it did was identify the spear that wouldn't break

𖤍

Altamash Khan

Mar 20, 20253 min read

I like AI, coding and tech. Contact me at altukhan43@gmail.com. Also, check out my siteand GitHub

Work Experience

Moneygram

Bengaluru, India
Senior Software Engineer (ML), Mar 2023 - Current

  • Worked on setting up Kubeflow pipelines with GitLab integration for the team
  • Worked on several document intelligence projects such as KYC verification using VLMs, data extraction using text segmentation, detection and recognition models
  • Developed a multilingual chatbot using Google’s Dialogflow CX, Natural Language AI, and Botcopy API
  • Integrated several features in the chatbot: live agent handoffs, authentication, Qualtrics Survey, transaction query, amend, and refund support
  • Integrated LLMs for chat summary; RAG over company FAQ docs
  • Used DOC AI and BERT QA models for document question answering to automate agent onboarding

Paytm

Bengaluru, India
Senior Software Engineer (ML), Dec 2022 - Mar 2023

  • Developed the ML orchestration framework using Azkaban and AWS EMR
  • Improved performance of models through better feature engineering and handling class imbalances
  • Implemented transient EMR cluster design to reduce costs and bottlenecks for Spark jobs

Capgemini

Bengaluru, India
ML Tech Lead, Apr 2021 - Dec 2022

  • Applied state-of-the-art deep learning models (CNNs, YOLO series, MMAction, VAEs, MS-Simple Baselines, GANs, Vision Transformers, etc.) to detect several unsafe events (person out of bounds / standing up, anomalous objects, person in distress) for Disney parks
  • Integrated MLOps (from data versioning to model registration) using AWS Sagemaker, Lambda, S3, and GitHub CI/CD for seamless retraining. Added a Drift monitoring framework built using AWS (Cloudwatch agent, Lambda, Athena, QuickSight) and Kubernetes
  • Optimized resource allocation in Kubernetes and model compilation to TensorRT, resulting in successfully handling 48 cameras running at 12 fps (every day), supporting 5 DL models in parallel with only 12 GPUs and an end-to-end runtime under 3 seconds
  • Successfully worked on a POC on NVIDIA Deepstream and TLT for tracking crowd density in theme parks (COVID times) and movement across regions of interest

Capgemini

Bengaluru, India
Data Engineer, Oct 2019 - Mar 2021

  • Designed and developed a new Data Lake python framework, with a test-driven development style, featuring ETL capabilities (using PySpark), job-recovery, job-control and data protection
  • Worked on a large scale Data migration project, covering over 50 subject areas, 15000+ tables, and 1000+ ETL processes, migrating from Teradata, Informatica and Maestro to Snowflake, AWS and UC4
  • Designed and developed cross-industry accelerators including TeraSnow (Python+Flask webapp for SQL conversion) and CGAttest (data validation tool with Tableau integration)

Education

NIT Silchar

Assam, India
M.Tech, Computer Science and Engineering

Projects

  • Real Time video analytics recipe, implementing small and large scale object detection from modeling to deployment
  • Dota2 games analysis using PySpark for insights on character picks and matchups
  • More projects available at: GitHub

Backlinks

  • No backlinks found