Sankho Roy

CEO

Hyderabad, Telangana, India9 yrs experience
AI ML PractitionerAI Enabled

Key Highlights

  • Expert in deep learning and AI automation.
  • Led successful AI projects in finance and geospatial domains.
  • Mentored teams and shaped AI strategies.
Stackforce AI infers this person is a Data Science and AI Automation expert with a focus on Finance and Geospatial Technology.

Contact

Skills

Core Skills

Deep LearningAi AutomationAi ConsultingAi StrategyTeam LeadershipGraph Neural NetworksAi SurveillanceMachine LearningAutomationFinancial ModelingNlp

Other Skills

Ideas DevelopmentLLMsAgentic AI workflowsPython ML ecosystemDeep Reinforcement LearningInformation RetrievalProbabilistic ProgrammingEmbedding methodsSpectral ClusteringApplied Machine LearningOCRComputer VisionClustering AlgorithmsAutoencodersHiring

About

Sankho Roy is a Lead Data Scientist with 8+ years of experience working across machine learning, deep learning, and applied AI research. His work focuses on designing advanced learning systems to solve complex problems in domains such as finance, automation, information retrieval, and computer vision. He is currently working on AI-driven automation for large-scale map operations, applying deep learning methods along with emerging approaches such as large language models and agentic AI workflows to improve operational intelligence and decision systems. Over the course of his career, he has worked across a wide range of AI domains including graph-based deep learning, reinforcement learning, probabilistic programming, evolutionary algorithms, computer vision, natural language processing, and information retrieval systems. His work often involves experimenting with different model architectures and learning paradigms to derive insights from complex datasets and automate knowledge extraction tasks. His areas of expertise include: • Deep Learning (CNN, RNN, LSTM, Transformers) • Graph Neural Networks and Graph-based Learning • Reinforcement Learning and Evolutionary Algorithms • Probabilistic Machine Learning and Bayesian Models • Natural Language Processing and Information Retrieval • Computer Vision and automated data extraction systems • LLM applications and agentic AI workflows He has also contributed to mentoring data scientists, shaping technical direction for AI initiatives, and developing research-driven solutions to real-world problems. Sankho completed his B.Tech from IIT Guwahati and has worked across consulting and enterprise environments applying advanced machine learning and deep learning techniques. He is particularly interested in the development of novel learning systems, deep learning architectures, and emerging AI paradigms combining neural models with autonomous agent systems. Contact: Email: sankhoroy@gmail.com Website: https://sankhoroy.github.io/

Experience

9 yrs
Total Experience
3 yrs 8 mos
Average Tenure
1 yr 8 mos
Current Experience

Apple

Lead Data Scientist at Apple Maps(Vendor via Thoughtgenesis)

Sep 2024Present · 1 yr 8 mos · Hyderabad · Hybrid

  • Key Contributions
  • Designing and building automation pipelines for Apple Maps operations using agentic AI workflows, LLMs, and deep learning systems.
  • Developing LLM-driven automation and decision pipelines to optimize large-scale geospatial data operations.
  • Building scalable AI systems integrating machine learning, deep learning, and generative AI models.
  • Collaborating with engineering teams to improve efficiency of maps data processing workflows.
  • Tools & Technologies
  • LLMs, Agentic AI workflows, Deep Learning, Python ML ecosystem, Automation pipelines
Deep LearningIdeas DevelopmentAI Automation

Decimal point analytics

3 roles

AI Consultant

Aug 2024Sep 2024 · 1 mo · Mumbai, Maharashtra, India · Remote

Deep Reinforcement LearningInformation RetrievalAI Consulting

Senior Data Scientist

Promoted

Feb 2021Jun 2024 · 3 yrs 4 mos

  • Leadership & Strategy
  • Managed and mentored a team of data scientists across multiple AI projects.
  • Led project roadmap planning, AI strategy, hiring, and technical mentorship.
  • Key Projects
  • Developed AI-based road-network planning models for India using Graph Neural Networks and probabilistic programming.
  • Built automation pipelines that significantly reduced manual effort in market research using LLMs and deep learning.
  • Designed an AI surveillance system to detect colluding groups in the Indian stock market using embedding methods and spectral clustering on 8M+ daily trades and orders.
  • Tools & Technologies
  • PyTorch, PyTorch Geometric, Uber Pyro, LLMs, RAG, Graph Neural Networks, Spectral Clustering, Deep Learning, Genetic Algorithms
Probabilistic ProgrammingDeep Reinforcement LearningAI StrategyTeam Leadership

Data Scientist

Jun 2019Feb 2021 · 1 yr 8 mos

  • Key Projects
  • Built an automated scanned-table-to-Excel system reducing manual work by 94% (3 hours → 10 minutes) using OCR, computer vision, and clustering algorithms.
  • Developed pair-trading insight models using autoencoder-based correlation discovery across financial instruments.
  • Conducted R&D on option pricing and asset allocation using reinforcement learning, probabilistic programming, and evolutionary algorithms.
  • Built an internal document retrieval system using BM25 search architecture.
  • Reduced manual work by 80% by building an automatic metric mapping system using KNN and BERT.
  • Built machine learning systems for loan approval prediction (95% accuracy) and NPA/bad loan detection (97% accuracy).
  • Developed a transaction tag classifier for 250k daily transactions, improving runtime performance from 1 hour to 20 seconds.
  • Tools & Technologies
  • PyTorch, TensorFlow, BERT, BM25, VAE, RBM, Autoencoders, Reinforcement Learning, Evolutionary Algorithms (CMA-ES, PEPG, Genetic Algorithms), KNN, Python ML ecosystem
Deep LearningApplied Machine LearningMachine Learning

Cgi

Machine Learning Engineer at CGI

Jan 2017May 2019 · 2 yrs 4 mos · Bengaluru Area, India · On-site

  • Built AI-based contract analysis system to extract key clauses and summarize legal documents using NLP.
  • Developed deep learning models for contract similarity and remediation recommendation using CNN-RNN attention and Siamese LSTM architectures.
  • Implemented algorithms for document comparison and clause matching including Gale-Shapley matching and LCS.
  • Built employee feedback summarization system using LSA/LSI based NLP techniques.
  • Developed computer vision model for automated damaged vehicle inspection for insurance claim processing using Inception V3.
  • Project selected for CGI ICE Innovation Challenge.
  • Tools: NLP, NLTK, Gensim, TensorFlow, CNN-RNN, Siamese LSTM, Computer Vision, Inception V3
Deep LearningApplied Machine LearningMachine Learning

Education

Indian Institute of Technology, Guwahati

Bachelor of Technology (BTech) — Chemical Technology/Technician

Jan 2013Jan 2017

Stackforce found 100+ more professionals with Deep Learning & Ai Automation

Explore similar profiles based on matching skills and experience