Solutions Architect Engineer · Columbus, Ohio

Anshuman Ranjan

I build innovative solutions leveraging multiple disciplines of computer science and engineering.

The Ohio State University logo

The Ohio State University

B.S. Computer Science & Engineering

Graduating December 2026

Amazon Web Services logo

Certification

AWS Certified Cloud Practitioner

01 / 07Applied Machine Learning Research

ML RESEARCH · OSU COLLEGE OF ENGINEERING· August 2026 – December 2026

Voice Cloning for Assistive Hearing

Benchmarking zero-shot voice cloning for hearing devices.

Undergraduate research at OSU's College of Engineering evaluating four zero-shot voice cloning architectures — F5-TTS, CosyVoice 2, XTTS-v2 and OpenVoice V2 — for deployment on assistive hearing hardware. Built speech-processing pipelines that measure model intelligibility across diverse signal-to-noise ratios.

PythonPyTorchHugging Face

4

zero-shot architectures evaluated

SNR

intelligibility tested across noise levels

So a hearing device can speak in a voice its wearer already knows.

Illustration of an ear with sound waves and a cloning waveform
02 / 07Full-Stack Software Development

SOFTWARE ENGINEER INTERN· May 2026 – August 2026

OCLC

Library cards, in your pocket.

Designed and shipped an end-to-end Google and Apple Wallet integration for CapiraMobile. A single codebase creates library-specific passes using each library’s branding and patron data, then signs and delivers them to Android and iOS.

Each pass pulls a library’s branding and patron fields, is cryptographically signed, and reaches the right Android or iOS app. Librarians control the exposed fields; no patron PII is stored in Capira’s databases.

JavaSpringKotlinApple WalletGoogle WalletDocker

200+

libraries supported by CapiraMobile

15,000

passes generated

Shipped to production in about three months, with accessibility, analytics and error logging built in.

Illustration of a library with a phone showing a wallet pass
03 / 07Embedded Networking Development

EMBEDDED NETWORKING ENGINEER· February 2026 – June 2026

Vision House Project

Security that runs on sunshine.

Built an off-grid AI security camera system for a non-profit: three solar-powered cameras, YOLO detection benchmarked across Raspberry Pi, Jetson and RTX 3060, streamed over WebRTC, RTSP and UDP.

PythonYOLORaspberry PiNVIDIADockerGStreamer

3

solar cameras

100%

off-grid uptime target

Surveillance that never sees a power bill — or the power grid.

Illustration of solar panels and a camera on a pole in a rural field
04 / 07Machine Learning Development

2ND PLACE · 24-HOUR HACKATHON· 2026

SCOTUSPredict

Predicting the Supreme Court.

A deep-learning cross-attention model over ModernBERT embeddings that forecasts case outcomes — placed 2nd at a 24-hour hackathon with 65% prediction accuracy.

PythonPyTorchHugging Facescikit-learn

65%

outcome accuracy

2nd

of all teams

Statutes in, probabilities out — ML applied to the highest court.

Illustration of a courthouse, gavel and probability bars
05 / 07Cybersecurity & IoT Security

SECURITY RESEARCH· 2026

Passive IoT Vulnerability Audit

Listening for weak spots.

A passive audit of IoT devices: Wireshark and PyShark capture traffic without touching the network, Hugging Face models turn findings into reports, Ghidra triages firmware.

PythonWiresharkHugging FaceBash

Passive

zero network disruption

LLM

generated reports

Found exposures in devices people never think twice about.

Illustration of a radar sweep with connected IoT devices
06 / 07Data Engineering & Visualization

DATA VISUALIZATION · 12-HOUR HACKATHON· 2025

OSU Energy Usage

A whole campus, by the watt.

Turned raw utility data into a living map of Ohio State: Pandas pipelines and Google Geocoding plotted consumption across 100+ campus buildings.

PythonpandasGoogle MapsLeafletReact

100+

buildings mapped

12 hr

concept to demo

Energy you can see — every building, one map.

Illustration of a campus skyline with power lines and a heat map
07 / 07Embedded Systems & Computer Vision

STARTUP · FOUNDER-BUILT

EspialNVR

A cloud-free NVR appliance with on-device AI.

A private network video recorder built on Raspberry Pi 5 with a Hailo-10H AI accelerator. Sub-second WebRTC live streaming, continuous local recording via Frigate, on-device facial recognition (SCRFD + ArcFace), automatic ONVIF camera discovery, and UPS battery telemetry — no cloud account, no subscription.

Raspberry PiPythonDockerHome AssistantWebRTCMQTT

<1s

live stream latency over WebRTC

0

cloud uploads — all AI runs on-device

Auto-deploys from a prod branch straight to the Pi — streaming, recording and recognition all stay local.

Illustration of a house with security cameras and a recording dashboard

Get in touch

Let’s talk about what we can build.

Open to full-time engineering roles after graduating in December 2026.

Tools & disciplines

Languages

PythonJavaKotlinSwiftC++CJavaScriptSQLBash

AI & ML

PyTorchTensorFlowHugging Facescikit-learnpandasNumPyLangChainClaude

Full-stack

ReactRuby on RailsFastAPIPostgreSQLElasticsearchTypeScript

Infrastructure

DockerKubernetesGoogle CloudDatabricksJupyterGitLinux