I am Puranjay Mishra
Name: Puranjay Mishra
Profile: SDE & ML Engineer
Email: puranjaymishra0@gmail.com
Phone: +1 716 709 0974
Tech stack:
Capability scores:
Core stack:
About me
I am a graduate student in Computer Science at SUNY Buffalo, specializing in AI/ML.
My foundation comes from Electronics and Communication Engineering at NIT Jamshedpur (an Institute of National Importance), where I built a strong base in HDLs, embedded systems, and signal processing. This low-level expertise complements my work in scalable software engineering and machine learning.
Beyond academics, I actively contribute to open source. I work on Burn (a Rust deep learning framework) and Twenty (a TypeScript CRM platform), collaborating with global communities and pushing real systems into production.
- DMI Finance: Expanded datasets with low standard deviation using PyTorch-based simulation/eval pipelines for distribution-shift stress testing.
- Petavue: Reduced analyst review time; cut failed executions and raised workflow success via observability and backend system design following Domain Driven Design.
Resume
ML-focused software engineer shipping production systems with rigorous evaluation, reliability, and scalable cloud infrastructure.
Summary
Puranjay Mishra
Machine Learning Engineer with 2 years of experience designing and deploying production-grade ML systems. Skilled in feature engineering, LLM context engineering, RAG pipelines, and rigorous model evaluation with reproducible training and inference workflows.
- Buffalo, NY, USA
- +1 716 709 0974
- puranjaymishra0@gmail.com
Education
MS in Computer Science, AI/ML Specialization
2024 - Present
State University of New York at Buffalo
Courses: Data Intensive Computing, Machine Learning, Deep Learning, Computer Vision & Image Processing
B.Tech in Electronics & Communication Engineering
2019 - 2023
National Institute of Technology Jamshedpur (Institute of National Importance)
Courses: Embedded Systems, Computer Vision, VLSI, Image Processing, Control Systems
Certifications
Professional Experience
Generative AI Engineer
Apr 2024 – Aug 2024
DMI Finance, Remote (Mumbai)
- Engineered an econometric-inspired artificial data generator with PyTorch; built scalable simulation + evaluation pipelines to stress-test models under distribution shift with low standard deviation outputs.
- Ran multi-node training on AWS EC2 with reproducible evaluation checks and runtime profiling, improving iteration speed for downstream experiments.
- Demoed ML pipeline to MUFG and a major U.S. bank, supporting FinTech investment validation and risk assessment.
Software Engineer
Jul 2023 – Apr 2024
Petavue, Chennai
- Built an NLP-focused AI explainer pipeline combining predictive modeling, LLMs & RAG to extract findings from unstructured documents (call transcripts, logs, documentation), improving search accuracy and reducing analyst review time.
- Designed a scalable RAG orchestration layer using similarity scoring and AWS S3-backed corpora, provisioned via Terraform, enabling reliable retrieval across large multi-tenant document sets.
- Shipped evaluation loops using Precision/Recall/F1/AUC and ablation-style testing to validate improvements and prevent regressions across model-backed workflows.
- Strengthened microservices using Domain Driven Design (DDD), SOLID principles, MVC design, and observability (Prometheus/Grafana), cutting failed executions and raising workflow success for real-time ML services.
Machine Learning Researcher
Dec 2022 – Jul 2023
Technical University of Crete, Greece
- Designed EEG-based time-series models with feature extraction and encoding pipelines, achieving 91.36% accuracy on multi-class prediction; evaluated using cross-validation and signal-level metrics.
- Developed scalable ML workflows for distributed training in the Digital Signal Processing Lab, focusing on reproducibility and experimental best practices.
Skillset highlights
Building reliable systems, from training loops to production services.
Systems Programming
Kernel-level work with interrupts, scheduling, memory management, and synchronization. Pintos OS project with 70+ kernel tests passed (Dockerized).
Full Stack Engineering
Microservices and product delivery using Spring Boot, React/Next.js, FastAPI, and REST APIs with CI/CD, testing, and release readiness baked in.
Distributed Systems
Deployed scalable services on AWS with Kubernetes and Terraform. Focused on cost, throughput, observability, and production reliability.
Machine Learning & LLM Systems
Production ML workflows with feature engineering, RAG pipelines, context engineering, and evaluation loops (Precision/Recall/F1/AUC; ranking metrics like nDCG@5, MRR).
AI Infrastructure
Training and inference on AWS (EC2/SageMaker), reproducible checks, profiling, CI validation, and monitoring with Prometheus/Grafana/CloudWatch.
Research Engineering
Peer-reviewed work and applied research across EEG time-series modeling, representation learning, and rigorous evaluation with reproducibility-first workflows.
Lead Author, IEEE BIBE
Funded Project @ ARTPARK, IISc
Undergrad Research Scholar, TUC
Top 1.4% – Joint Entrance Exam 2019 (1.1M+)
Publications
Peer-reviewed work and ongoing research.
Colour Prediction using Vision Transformer and Continuous Wavelet Transform on EEG Signals
IEEE BIBE 2023
Lead author • DOI: 10.1109/BIBE60311.2023.00036
Cyphersense: User Preference Modeling for News Recommender Systems
In progress
Targeting submission in 2026 • Human feedback + preference modeling for large-scale personalization and alignment
Portfolio
Selected work across apps, ML systems, and research.
- All
- App
- Research
- Blogs
Promptline
A DevOps/MLOps proof-of-concept co-built with an industry veteran: an AI agent that safely versions, modifies, and validates infrastructure/config changes with auditable automation, rollback-minded workflows, and CI/CD triggers via serverless control flows. Includes JUnit validation tests and shell contract probes. Tech: Spring AI, AWS Lambda/EC2, GitHub Actions, Next.js.
UnCypher
A hyperpersonalized navigator with a navigation-focused ML attribution pipeline (FAISS + HDBSCAN) and streamed contextual insights. Evaluated and optimized ranking quality using nDCG@5 and MRR, achieving SOTA performance on the Meta News dataset via improved scoring and retrieval strategies.
Contact
Address
Buffalo, New York
Call at:
+1 716 709 0974
Email at:
puranjaymishra0@gmail.com