➜ ~ whoami

Vistrit Pandey

const role = ""; AI Engineer working on applied AI, ML systems and inference, and forward-deployed engineering.

I build production AI systems end to end — model evaluation, local inference infrastructure, agentic and multimodal pipelines, and the backend and product surfaces that put them in front of real users.

get in touch resume.pdf

Chicago, ILSoftware Engineer II, AI R&D @ Blinkfire Analytics

Vistrit Pandey, AI engineer based in Chicago — Applied AIML Systems & InferenceForward-Deployed Engineering

01 — experience.log

Where I've worked

Four years of shipping AI systems — from research prototypes to services customers depend on.

Software Engineer II, AI R&D

Blinkfire Analytics

Jan 2026 — PresentChicago, IL● running

// Building production AI systems end to end across model evaluation, inference, backend services, APIs, and customer-facing applications.

→Build production AI systems end to end across model evaluation, inference, backend services, APIs, and customer-facing applications in Python, Go, TypeScript, Next.js, and Google Cloud.
→Design and operate local LLM inference infrastructure with CUDA, TensorRT, NVIDIA Triton, Ollama, Docker, Kubernetes, and Linux — optimizing model serving, latency, throughput, deployment control, and cost.
→Develop reinforcement learning, synthetic data generation, fine-tuning, and evaluation pipelines over 12+ years of proprietary multimodal data for LLM training and benchmarking.
→Ship agentic and multimodal systems combining RAG, semantic search, tool execution, Whisper, LLMs, and vision-language models for analytics, automation, audio understanding, and structured outputs.
→Partner with product, infrastructure, leadership, and enterprise stakeholders to take AI capabilities from research prototypes into production services and customer workflows.

Software Engineer I, AI R&D

Blinkfire Analytics

Apr 2024 — Dec 2025Chicago, ILexited 0

// Shipped full-stack AI and analytics products across distributed service architectures.

→Shipped full-stack AI and analytics products using Python, Go, TypeScript, REST, gRPC, Redis, Next.js, Vue.js, Firebase, and GCP across distributed service architectures.
→Integrated commercial and open-weight models into semantic search, content intelligence, automation, and real-time inference; built fine-tuning, quantization, serving, and inference optimization workflows with PyTorch, TensorFlow, CUDA, TensorRT, and Triton.
→Built backend APIs, asynchronous processing systems, annotation tooling, and customer-facing interfaces supporting production AI workflows.

Founder and Software Engineer

Meetext

Mar 2023 — Mar 2024Remoteexited 0

// Independent project — founded and shipped an AI communication platform to 20,000+ users.

→Founded and built an AI communication platform with Next.js, Python, Firebase, speech recognition, LLM APIs, and Stripe, scaling it to more than 20,000 users.
→Led seven engineers and designers while owning system architecture, cloud deployment, payments, product delivery, user support, and technical execution from prototype through public launch.

Course Assistant, Computer Science

Drexel University

Sept 2022 — Sept 2023Philadelphia, PAexited 0

// Supported computer science coursework through mentorship, labs, and evaluations.

→Mentored 500+ students across algorithms, machine learning, computer vision, computational imaging, programming, and software engineering through labs, debugging, code reviews, and project evaluations.

02 — projects.tsx

Things I've built

Products, infrastructure, and research I own end to end.

MachBoost

Open-source local AI infrastructure

open source

A native macOS app, Python package, and team inference gateway for resident MLX text and vision models — authenticated LAN serving, load-aware routing, bounded queues, and OpenAI-, Anthropic-, and Ollama-compatible APIs. Repository-aware prefix reuse, multimodal caching, and acceleration paths measured a 2.89× median speedup across token-identical repeated repository workloads under paired benchmarks.

MLXPythonSwiftmacOSInference GatewayBenchmarking

machboost.xyz source

Meetext

Forward-deployed engineering platform

platform

A deployment control plane for shipping AI products, MCP servers, and APIs into customer environments — OpenAPI-to-MCP conversion, per-customer capability isolation, versioned deployments, and multi-destination delivery. Includes OAuth and permission validation, security review gates, continuous health checks, incident diagnostics, credential lifecycle management, and end-to-end deployment verification.

Next.jsPythonMCPOpenAPIOAuthDeployments

meetext.xyz

MeetHealth

Applied AI and model evaluation

product

An AI platform for dietitians supporting 200+ organizations, backed by 1M+ locally generated synthetic clinical diet plans used for guideline-anchored evaluation, training data, and domain-specific model development.

Next.jsPythonGoogle CloudCUDALocal LLM InferenceSynthetic Data

meethealth.us

Grounded Introspection

Model interpretability research

research

A white-box framework testing whether an LLM's verbal uncertainty tracks its internal next-token entropy — read-only hidden-state and logit access, bounded autonomous generation, and permutation controls. Reproducible code, raw step-level records, and a working paper from a preliminary Qwen3 pilot, explicitly treated as underpowered.

MLXPythonInterpretabilityEvaluation

github.com/VistritPandey/grounded-introspection

03 — stack.json

What I work with

The tools I reach for across models, systems, and product.

stack.json35 entries

{
  "languages": ["Python", "Go", "TypeScript", "JavaScript", "SQL", "Java", "C++", "Bash"],
  "ai_and_ml": ["PyTorch", "TensorFlow", "Hugging Face", "MLX", "Reinforcement Learning", "Fine-tuning", "Evaluation", "Synthetic Data", "Quantization", "RAG", "Multimodal AI", "Whisper", "VLMs"],
  "systems_and_product": ["CUDA", "TensorRT", "Triton", "Ollama", "Docker", "Kubernetes", "Linux", "GCP", "AWS", "Redis", "gRPC", "React", "Next.js", "Firebase"]
}

04 — education.yml

Where I studied

University of Illinois Urbana-Champaign

degree: Master of Computer Science
dates: 2026 — 2027 (Expected)
location: Illinois, US
mode: Online

Drexel University

degree: B.S. Computer Science, Minor in Information Systems
dates: 2019 — 2023
location: Pennsylvania, US

05 — contact.sh

Let's build something worth shipping

Applied research, AI infrastructure, or a product from zero to one — I'd like to hear about it.

zsh — contact

➜ ~ mail vistrit2@illinois.edu

vistrit2@illinois.edu

➜ ~ open github

github.com/VistritPandey

➜ ~ open linkedin

www.linkedin.com/in/vistrit

➜ ~ curl -O resume.pdf

Resume.pdf — 114 KB

➜ ~