Software Developer, Oracle
Interests: Deep Learning & Machine Learning · Distributed Systems · Agentic Systems
Education
| Year | Degree | Institute | CGPA |
|---|---|---|---|
| Aug 2020 - May 2024 | B.Tech, Computer Science | Indian Institute of Technology (IIT) Bombay | 8.31/10 |
Teaching Assistant – IIT Bombay
- PH107: Quantum Physics and Applications
- PH108: Electricity and Magnetism
- Assisted course instruction and provided academic support for 45+ undergraduate students.
- Graded assignments, answered queries, and contributed to exam preparation and course material.
Work Experience
Software Developer (IC2) | Oracle – MicroTx Agentic Workflows
Jul 2024 - Present
- Evaluated workflow execution engines (Conductor, Temporal, Restate) and led adoption of an open-source orchestrator; analyzed consistency, failure recovery, and scalability trade-offs.
- Extended the orchestration platform with AI-native tasks and service connectors, enabling LLM-driven agentic workflows coordinating long-running microservice operations.
- Implemented durable workflow execution using persistent state storage, retries, compensation (saga-style recovery), and idempotent task handling across distributed services.
- Currently hardening the platform with a focus on security and performance to improve reliability, low-latency scheduling, and high-throughput task processing before release.
Internships
Software Developer Intern | Oracle – MicroTx Distributed Transactions
Summer 2023
- Designed and integrated LRA / Saga patterns for reliable distributed transaction management.
- Developed SDKs for distributed transaction protocols: XA (two-phase commit), LRA (long-running transactions), and TCC (try-confirm-cancel), with native Spring Boot integration.
Software Developer | HeltarAI
Summer 2022
- Built a customer communication platform integrating Meta WhatsApp Cloud Graph API for automated workflows.
- Developed real-time bidirectional messaging using WebSockets for low-latency client-server communication.
Key Projects
Language Modeling from Scratch (CS336)
- Built foundational language modeling components from the ground up, implementing Byte-Pair Encoding (BPE) tokenization and autoregressive transformer blocks.
- Learning low-level GPU programming for writing custom kernels in CUDA, triton
Deep Learning for Computer Vision (CS231n)
- Built fundamental machine learning classifiers (k-NN, SVM, Softmax) and a two-layer neural network from scratch using NumPy on CIFAR-10.
- Implemented modular deep learning components with custom forward and backward passes for CNNs, Batch Normalization, and Dropout.
- Developed an image captioning system using RNNs and LSTMs, and implemented methods including SimCLR, CLIP, DINO, and DDPM.
Natural Language Processing with Deep Learning (CS224n)
- Analyzed and visualized Word2Vec and GloVe embeddings, evaluating semantic and syntactic structure through similarity and analogy probes.
- Developed a sequence-to-sequence Neural Machine Translation (NMT) model using an RNN encoder-decoder with multiplicative attention.
- Implemented a Transformer model from scratch, including multi-head self-attention, positional encoding, and feed-forward blocks.
Adversarial ML: Evasion Attacks on Classifiers
B.Tech Research Project – Advisor: Prof. Virendra Singh (Spring 2024)
- Studied adversarial ML in security-critical settings.
- Implemented gradient-based attacks against linear models and SVM classifiers for error-generic and error-specific evasion.
- Evaluated on MNIST and explored defenses such as adversarial training.
Multi-Language Question Answering System
Guide: Prof. Pushpak Bhattacharya (Fall 2023)
- Built a multilingual extractive QA system supporting Indian languages.
- Fine-tuned BERT on SQuAD (100k+ question-answer pairs across 500+ Wikipedia articles).
- Achieved strong evaluation results with an 80% Exact Match and 88% F1 score.
Eye Health Monitor – Low-Cost Wearable Device
Koita Center for Digital Health, IIT Bombay (Fall 2023)
- Prototyped wearable using ESP32-CAM and macro lens to capture ocular images.
- Collected and curated datasets for cataracts, glaucoma, uveitis, and related conditions.
- Designed CNN and Vision Transformer (ViT) architectures for automated eye disease classification.
Scholastic Achievements
- AIR 199 – JEE Advanced
- AIR 497 – JEE Main
- AIR 40 – JEE Main B.Planning
- State Rank 18 – AP EAMCET
- State Rank 35 – TS EAMCET
- Top 450 – Indian National Physics Olympiad (INPhO)
- Top 330 – Indian National Astronomy Olympiad (INAO)
- Top 800 – Indian National Chemistry Olympiad (INChO)
Patents
- Native Integration of Retrieval-Augmented Generation (RAG) pipelines in durable workflows
- System and Method for Agentic Task Execution in MicroTx Workflow Engine
- Multi-Agent Orchestration with Dynamic Workflow Composition using Agentic planner in workflow
- Transaction System Task for Distributed Transactions in MicroTx Workflow
Last updated: 3 Mar, 2026