Aditya Mhaske

Software Engineer & Researcher

Specializing in Multi-Agent Orchestration, LLM Evaluation Harnesses, and Multimodal Edge/Cloud AI Architecture. Designing robust agent topologies with structured tracing, human-in-the-loop DAG execution, and production RAG systems.

Animal Coder Sticker (Day Mode) Animal Coder Need Coffee Sticker (Dark Mode)
Focus Areas: Multi-Agent DAGs & Orchestration LLM Evaluation & Tracing Multimodal Signal Processing RAG & Vector Databases Edge / Cloud AI Systems
02

GitHub Contributions

Loading contribution data...
03

All Repositories & Systems

Showing 14 systems
System / Repository Architecture & Summary Tech Stack Actions
Open-source multi-agent research platform with Planner, Researcher, Critic, and Synthesizer agents, human-in-the-loop validation, and per-claim citations.
LangGraph·FastAPI·Next.js
EVALS & TRACING agent-arena
Evaluation harness that ranks models and whole multi-agent pipelines on your own accuracy, cost and latency criteria, plus a frozen study benchmarking four coordination topologies with structured telemetry.
LLM Evals·Structured Tracing·Python
Natural-language AI co-pilot translating conversational analytics queries into structured backend API calls via Anthropic function calling.
Anthropic API·FastAPI·Python
AGENTIC AUTOMATION computer-use-automation
LLM-driven automation for legacy apps with no API: discovers a workflow once, compiles it into a typed capability artifact, then replays it deterministically with no model in the loop.
Playwright·LLM Agents·Python
INDUSTRIAL RAG IndusAI
Industrial AI assistant for PLC fault log analysis and diagnostics; architected GCP cloud migration with sub-second vector matching.
Gemini Flash·Qdrant·Cloud Run
Data science and NLP research with Kelley School of Business, analyzing political Twitter text, customer intent models, and cross-classified tracking.
LLM / NLP·Python·Jupyter
MULTIMODAL AI AffectiSense
NeuroSense AI — Multimodal mental health assessment platform fusing neurophysiological (EEG), vocal acoustic biomarker, and facial signals.
PyTorch·EEG Processing·Python
Multimodal behavioral analysis platform combining facial micro-expression telemetry and acoustic signal processing for mental health markers.
Multimodal AI·Signal Proc·Python
DATA PIPELINE BMR-ML-Pipeline_DBMS
End-to-end ML data pipelines, relational database architecture in Oracle SQL, and business intelligence dashboards in Tableau.
Oracle SQL·Python·Tableau
Supervised machine learning pipeline evaluating classification model ensembles for high-sensitivity cardiovascular disease risk prediction.
Scikit-Learn·Python·Jupyter
Distributed data preprocessing, wrangling, and financial forecasting pipeline for Deutsche Börse stock data deployed on AWS.
AWS·Time-Series ML·Python
Spatial clustering and location recommendation system using Foursquare venue telemetry and exploratory geospatial data analysis.
Geospatial ML·Python·Jupyter
FULL STACK Medical-Dashboard
Modular hospital management system and medical dashboard providing central administrative tools for doctors, patients, and clinical staff.
TypeScript·React·Node.js
IOT & EMBEDDED Green-house
Embedded IoT greenhouse monitoring system integrating gas detectors, temperature/light sensors, water level probes, and an LCD warning panel.
C++·Arduino·Sensors
04

Patents & Peer-Reviewed Publications

Google Scholar Profile
[01] Patent

Smart Glasses for Monocular Vision to Estimate Depth

Patent Number: IN 202021034153 · Indian Patent Office · Filed Aug 10, 2020

Patented wearable assistive system combining monocular vision feed processing and disparity estimation to reconstruct real-time depth cues.

[02] Patent

Road Conditions and Obstacle Indication System

Patent Number: IN 202021009028 · Indian Patent Office · Filed Mar 3, 2020

Patented autonomous vehicle perception architecture for real-time roadway anomaly detection, obstacle localization, and automated warning indication.

[03] IEEE Conference

EEG Based Depression Detection Using Ensemble Approach

International Conference on Applied Intelligence and Sustainable Computing (ICAISC)

Developed ensemble machine learning algorithms on raw EEG time-series signals to classify neurophysiological biomarkers of depression.

[04] IEEE Conference

A Survey: Feature Extraction Techniques And ML Models For Depression Analysis

12th Intl. Conference on Computing, Communication, and Networking Technologies (ICCCNT)

Comprehensive review and comparative benchmarking of multimodal feature extraction techniques across speech, facial micro-expressions, and physiological signals.

[05] IEEE Conference

Cardiovascular Disease Prediction Using Machine Learning Models

IEEE Pune Section International Conference

Comparative evaluation and optimization of supervised ML models for high-sensitivity cardiovascular risk factor prediction.

[06] IEEE Conference

Road Conditions Indication and Autonomous Braking System

IEEE 3rd International Conference on Intelligent Sustainable Systems (ICISS)

Architected embedded edge system linking road surface computer vision telemetry directly to automated vehicle braking actuation.

[07] IEEE Conference

Smart Glasses to Assist Monocular Vision to Estimate Depth

IEEE International Conference for Innovation in Technology (InCACCT), Bangalore

Academic paper presenting mathematical depth estimation models and sensor fusion techniques implemented in the patented smart glasses.