Bhoomika R Hegde
About
I enjoy solving problems that don't come with obvious answers. My curiosity usually leads me to papers, prototypes, and plenty of experimentation.
I'm interested in research-driven engineering and building intelligent systems that solve meaningful problems. Currently an AI Engineer Intern at Yugasys, pursuing a B.Tech in Computer Science (AI & ML) at PES University.
Publications
-
2026
Discrete Representation Learning: A Review
Accepted -
2025
Transaction Graph-Based Predictive Hurdle Model for Credit Scoring in DeFi Lending Protocols
Under Review
Experience
-
AI Engineer Intern
- Built a voice AI orchestration backend integrating speech-to-text, intent parsing, and app-side actions across 60+ flows, enabling end-to-end voice-driven user interactions.
- Engineered real-time communication systems using WebSockets and Agora SDK with state management, exponential backoff, and heartbeat monitoring for reliability on unstable networks.
- Designed an agent-based integration layer using MCP with clear service boundaries and event contracts; applied domain-driven design and resolved failure modes across multi-service interactions while maintaining CI/CD stability.
-
AIML Engineer Intern
- Built a modular data processing pipeline (Python) spanning 5 blockchain protocols for bot detection, enabling protocol-specific preprocessing components to evolve independently.
- Implemented end-to-end traceability across the pipeline, tagging each transaction from ingestion through classification to ensure debuggability and auditability.
- Improved production readiness by reducing silent failure modes, adding validation checks at stage boundaries, and ensuring error states were logged and observable.
-
Teaching Assistant — Advanced Data Analytics
- Wrote exercises and case studies for a 250-student course; sat with student teams on architecture decisions and tricky debugging sessions.
- Mentored teams on project architecture, debugging strategies, and efficient algorithm implementation.
- Standardised grading rubrics to prevent assessment drift across 250 submissions.
-
Teaching Assistant
- Helped student teams build a blockchain-based learning platform with NFT badges and smart contracts focused on the seam between smart contract logic and the UI, where integration tends to break.
-
Engineer Intern
- Developed a Python-based ETL system to parse and normalise EPA documents into structured outputs using a modular plugin architecture for inconsistent formats.
- Integrated a multi-step reasoning pipeline using ReAct and AutoGen; added validation and logging at each stage to prevent silent data corruption.
- Collaborated with cross-functional teams, ensuring modular updates without breaking downstream pipeline components.
Education
-
B.Tech — Computer Science & Engineering (AI & ML)
PES University · Bengaluru
- DAC Scholarship recipient
- Head of Sponsorship, Aatmatrisha '25
- Head of Sponsorship, Chords '24
- Head of Social Media, AIKYA
Projects
-
Systems · ML · 2024
Dog Activity Recognition System
Multi-service, event-driven platform (Python, Kafka, PostgreSQL, FastAPI, Docker) with a unified session-ingest system time-aligning video frames with biosignal streams. Deployed a CLIP-based zero-shot inference service with audit logs, dead-letter queues, and retry mechanisms for reliability under failure.
-
ML · Research · 2024
Image Representation via ViT + Product Quantisation
Identified limitations in VQ-VAE2 and proposed PQ-VAE, combining Vision Transformers with structured latent decomposition and differentiable codebook learning. Improved reconstruction quality and compression efficiency over baseline, validated using SNR and SSIM. Companion paper accepted at ICMLC 2026.
-
Backend · Systems · 2024
Relational Knowledge Graph for Question Answering on Relational Databases
Developed a relational knowledge graph to enhance question answering capabilities on relational databases. Implemented efficient query processing and reasoning mechanisms to improve accuracy and performance.
-
Backend · Architecture · 2024
Academic Management System
Microservice-based university portal with clearly defined service boundaries and API contracts enabling independent deployment. Domain-driven design separating scheduling, feedback, and user management into distinct bounded contexts. Containerised with Docker.
No projects match this filter yet.
Skills
Languages
- Python
- C++
- SQL
- JavaScript
- Solidity
- Bash
AI / ML
- PyTorch
- Transformers
- ViT
- ReAct
- AutoGen
- CLIP
- LLM Pipelines
- Vector Quantisation
Systems & Infrastructure
- Apache Kafka
- FastAPI
- WebSockets
- Docker
- PostgreSQL
- REST APIs
- CI/CD
- Agora SDK
Practices
- Domain-Driven Design
- Event-Driven Architecture
- Microservices
- ETL Systems
- Observability
- API Contract Design
From LinkedIn
A recent post — for the full feed, visit my LinkedIn profile.
Contact
Open to research collaborations, internships, interesting engineering problems, and good conversations about systems. Best reached by email.
bhoomikahegde85@gmail.comI aim to reply within two working days.