VarunDronamraju
Track Record & Research Systems

Engineering & Industry Experience

Full technical career history across quantum communication software at IISER Bhopal, on-device ML architecture at Wellness at Work, and molecular pipeline engineering at QClairvoyance.

Founding Ventures

Looking for LogicKrew Ventures & Production AI Systems?

Explore MYCity, KrewAI, the Agent Execution Platform, and LocationHQ on the dedicated platform page.

View LogicKrew Systems→
EXPERIENCE & INSTITUTIONS

Career Trajectory & Track Record

Chronological engineering progression across quantum research, early-stage ventures, and production ML platforms.

Aug 2026 – PresentQuantum Simulation

IISER Bhopal

Full Stack Developer

Architecting quantum photonic network simulation platforms and cross-platform desktop execution pipelines for quantum communication systems.

PythonTypeScriptRustReactRedisDocker
July 2026 – PresentHyperlocal Geo Systems

LogicKrew (MYCity)

Founder & AI Engineer

Building location-aware discovery platform using Flutter, FastAPI, PostGIS, and XOS multi-city control plane.

FlutterFastAPIPostgreSQLPostGISRedisFirebase
Feb 2026 – PresentAgentic AI & Observability

LogicKrew : KrewAI

Founder & AI Engineer

Built production LLM observability, prompt tracing, pgvector memory stores, and agent orchestration pipelines.

FastAPIPostgreSQLpgvectorDockerRedisLLMs
Aug 2025 – Feb 2026Desktop ML & Health

Wellness at Work Pvt. Ltd.

Full Stack ML Engineer

Shipped Lumina AI Wellness for macOS and Windows with system-level ML features, real-time telemetry, and Firebase infrastructure.

ReactFirebasePythonDesktop OS APIs
Aug 2024 – Jun 2025Bioinformatics ML

QClairvoyance Quantum Labs

Machine Learning Engineer

Built automated ingestion pipelines for 600K+ molecular compounds and full-stack scientific data management platform.

ReactNode.jsPostgreSQLPython
2021 – 2025Education & Foundations

GITAM University

B.Tech in Computer Science Engineering (AI & ML)

Core foundations in algorithms, asymptotic analysis, deep learning, distributed systems, and machine learning.

PythonAlgorithmsC++ML
RESEARCH & INDUSTRY

Industry & Research Positions

Positions and research appointments spanning quantum network simulation, on-device computer vision ML, and molecular data pipelines.

Research & SystemsBhopal, India

Indian Institute of Science Education and Research (IISER) Bhopal

Full Stack Developer

Aug 2026 – Present
Problem Statement

Simulating quantum photonic network topologies and calculating quantum state fidelity under physical channel noise requires high-performance simulation pipelines with accessible desktop controls for quantum researchers.

Solution Design & Architecture

Architected an end-to-end desktop simulation suite using Python, TypeScript, and Rust, implementing quantum state propagation engines, entanglement distribution routing, Redis caching, and Dockerized execution pipelines.

Impact & Key Metrics
  • ✓Built cross-platform desktop execution pipeline simulating 50+ optical node configurations.
  • ✓Reduced simulation configuration latency from manual CLI scripting to one-click visual topology validation.
  • ✓Engineered automated telemetry export and state fidelity benchmarking with low-overhead Rust & Redis pipelines.
Stack:PythonTypeScriptRustReactRedisDocker
Full-Time IndustryRemote

Wellness at Work Pvt. Ltd.

Full Stack Machine Learning Engineer

Aug 2025 – Feb 2026
🏆#1 Product of the Week on LaunchPadIndia · 4.8 Rating on Mac App Store
Problem Statement

Remote professionals experience severe Computer Vision Syndrome (CVS) and posture fatigue from 8+ hour screen days, yet existing wellness tools rely on disruptive manual timers that users immediately dismiss.

Solution Design & Architecture

Architected Lumina AI Wellness as a native macOS & Windows application powered by low-overhead, on-device ML models. Senses eye strain, blink rate drop, and posture drift in real time without video leaving the user's device, triggering non-intrusive micro-nudges.

Impact & Key Metrics
  • ✓Adopted by 10,000+ remote professionals globally.
  • ✓Maintained 4.8 / 5.0 rating on the Mac App Store.
  • ✓Sub-2% CPU consumption with optimized on-device computer vision inferences.
  • ✓Reduced user-reported eye fatigue symptoms by 65% in active cohorts.
Stack:ReactFirebasePythonCoreMLDesktop OS APIsComputer Vision
Research LabHyderabad, India

QClairvoyance Quantum Labs

Machine Learning Engineer

Aug 2024 – Jun 2025
Problem Statement

Biochemical discovery teams faced severe friction querying fragmented, heterogeneous compound libraries spanning over half a million chemical structures, stalling computational drug discovery cycles.

Solution Design & Architecture

Designed and automated parallelized data ingestion pipelines to clean, normalize, and index 600K+ molecular compounds into PostgreSQL, complemented by a full-stack React and Node.js exploration workspace with molecular structure rendering.

Impact & Key Metrics
  • ✓Structured & ingested 600,000+ molecular compounds into a queryable relational database.
  • ✓Accelerated chemical property lookup latency by 10x across distributed research teams.
  • ✓Enabled ML teams to train molecular property prediction models directly from unified data streams.
Stack:ReactNode.jsPostgreSQLPythonRDKitData Pipelines
SCIENTIFIC ML

Research & Scientific Systems

De Novo drug discovery predictive models, high-precision behavioral profile classification, and algorithms.

De Novo Drug Discovery Model

Predictive ML on 900K+ Molecular Structures for Target Interaction

Scientific ML

Developed ML models on molecular datasets (900K+ compounds) to predict drug-target interactions and identify critical inhibitory features with structured data pipelines.

Problem Solved

Synthesizing drug candidates experimentally costs millions; computational screening requires accurate biological inhibitory feature predictions before laboratory synthesis.

Solution Architecture

Developed deep learning models on molecular datasets (900K+ compounds) using RDKit for chemical fingerprinting and TensorFlow for target interaction classification and inhibitory feature detection.

Impact & Key Metrics
  • ✓Trained on 900,000+ chemical compounds with reproducible feature extraction.
  • ✓Achieved high ROC-AUC on inhibitory affinity prediction across target proteins.
PythonTensorFlowRDKitScikit-learn

Behavioral Feature Fake Profile Detector

High-Precision Social Graph Analysis on 37.6K Profiles

Machine Learning

Classification system analyzing 37.6K+ social profiles, achieving 97.8% precision through behavioral feature engineering.

Problem Solved

Automated bot accounts and synthetic social identities bypass superficial text filters by mimicking superficial profile data.

Solution Architecture

Engineered behavioral and structural graph features across 37,600+ social profiles, training gradient-boosted and ensemble classifiers focusing on activity cadence and interaction density.

Impact & Key Metrics
  • ✓Achieved 97.8% precision on unseen test datasets.
  • ✓Benchmarked across 37,600+ profiles with robust feature ablation testing.
PythonScikit-learnPandasNumPy