About Me

I am a researcher and engineer passionate about building interpretable, reliable, and scalable machine learning systems.

My background uniquely combines rigorous academic research in Explainable AI (XAI) with hands-on experience engineering petabyte-scale streaming platforms.

Publications & Preprints

  • A. Prabhakaran, et al. “MorphoSHAP: Rethinking the Unit of Attribution in Explanation for Deep Visual Models.” Submitted to CVPR 2027.

  • A. Prabhakaran, et al. “Enhance Eye Disease Detection using Learnable Probabilistic Discrete Latents in ML Architectures.” 2025.

  • A. Prabhakaran, et al. “Leveraging Deep Learning for Fever Temperature Analysis and Pattern Recognition.” IEEE AICECS 2024. (Best Paper Award)

Education

Institut Polytechnique de Paris (IP Paris) | Palaiseau, France
M2 Machine Learning, Communications, and Security | Sept 2025 - Aug 2027

  • M1 (Data Analysis and Pattern Classification) GPA: 18.1/20 (Rank 1/25)
  • Coursework: Deep Learning, Foundations of Machine Learning, Convex Optimization, Sequential Decision-Making & RL.

National Institute of Technology Karnataka (NITK) | Surathkal, India
B.Tech in Electronics and Communication Engineering, Minor in CSE | Dec 2020 - May 2024

  • GPA: 9.16/10 (Major), 9.00/10 (Minor)
  • Undergraduate Thesis: Engineered the ExTemp architecture for multi-class physiological time-series classification, interpreting model decisions via GradCAM.

Applied Research Experience

XAI and Mathematical Morphology Research Intern | AMIAD & ENSTA, IP Paris
Nov 2025 - Present

  • Architected MorphoSHAP, an explainable AI framework redefining cooperative Shapley game attribution from raw pixels to hierarchical morphological structures.
  • Attained SOTA Insertion/Deletion AUC across 5 benchmarks while accelerating explanation latency by 6x over superpixel KernelSHAP, enabling real-time production model introspection.

AI Research Intern | CogAI4Sci Lab, National University of Singapore (NUS)
Aug 2023 - Jan 2024

  • Formulated a diagnostic framework using GFlowOut in Vision Transformers and ResNet18 for ocular disease classification.
  • Validated epistemic uncertainty estimates to monitor out-of-distribution drift, demonstrating model reliability under severe input distribution shifts.

Engineering Experience

Software Engineer 1 | Uber Technologies Ltd.
June 2024 - Aug 2025

  • Engineered distributed streaming infrastructure on Apache Flink, sustaining petabyte-scale real-time telemetry pipelines powering production ML feature ingestion and dynamic pricing.
  • Executed a zero-downtime infrastructure migration of 1,000+ distributed YARN jobs, preserving strict stateful checkpoint consistency.

Technical Skills

  • Languages & Frameworks: Python, PyTorch, C++, Java, Scala, SQL, Golang, Flink, Spark, Kafka, Docker.
  • Focus Areas: Applied Machine Learning, Model Observability & XAI, Large-Scale Distributed Systems, Biologically Plausible Learning, Uncertainty Estimation.