01
Firmware Validation
Core role · Toshiba
Stress-testing SMR & Hybrid-SMR SATA drives — long-running I/O and zone-operation workloads that surface firmware defects before they ever reach a customer.
Open to opportunitiesSoftware Engineer · ToshibaSilicon Valley · on-site
Keegan PaulColaco.
Where bits meet magnetism.
Software Engineer with 3+ years building firmware-validation and performance tooling for Toshiba's SMR and Hybrid-SMR hard drives — now pushing toward next-generation SSDs, with published deep-learning research and a working knowledge of modern AI on the side.
- 0.54
- CGPA — BTech CSE
- 0+
- Years — HDD firmware
- 0
- Springer paper
- 0
- Hackathon podiums
7,200RPM
C / C++PythonSMR HDD firmwareStress testingwxWidgetsLinuxGit / SVNFIO · VDBenchPerformance analysisYOLOv8Deep learningData visualisation日本語 · JLPT N5C / C++PythonSMR HDD firmwareStress testingwxWidgetsLinuxGit / SVNFIO · VDBenchPerformance analysisYOLOv8Deep learningData visualisation日本語 · JLPT N5
01About
An engineer at the lowest level.
Keegan is a Software Engineer in Toshiba's HDD division, where he's spent the last three-plus years on the unglamorous, high-stakes work that keeps your data safe: validating the firmware inside SMR and Hybrid-SMR hard drives, and building the tooling that proves it's ready to ship.
That means stress-testing drives for hours on end, turning raw failure dumps into something a developer can read, and benchmarking the performance metrics that define a drive's character. He's now setting his sights on next-generation SSDs.
He came up through SRM with a 9.54 CGPA and a software-engineering specialisation, picked up a few hackathon podiums along the way, and published deep-learning research under Springer. Off the clock he's tracking the AI landscape and chipping away at Japanese.
Spec sheet
REV · 2026- ROLE
- Software Engineer
- DIVISION
- Toshiba · HDD
- BASED
- Bengaluru, IN
- ORIGIN
- Goa, IN
- DEGREE
- BTech CSE · 9.54
- STACK
- C/C++ · Python
- LANGUAGES
- EN · 日本語 N5
STATUS · OPEN TO OPPORTUNITIES
02Expertise
What I'm good at.
From firmware that can't fail to models that can see — four areas where the work gets interesting.
02
Performance Engineering
FIO · VDBench
Benchmarking the metrics that matter, then building the CLI and web tools that parse, visualise and make sense of mountains of HDD performance data.
03
Debug Tooling
C++ · wxWidgets
A Windows GUI debugger that turns raw failure dumps pulled off a drive into something an engineer can actually read — cutting firmware debug time.
04
AI / Machine Learning
Published research
Peer-reviewed YOLOv8 work under Springer, plus a working grasp of neural nets, LLMs and RAG — applying deep learning to real-world detection problems.
03The platter · 3D
Inside the drive.
Keegan's day job lives at the lowest level of storage — where firmware meets spinning magnetic media. Here's that world, rendered in real time: drag to orbit, cut the power, toggle the track map, or fire a seek.
04Experience
The work.
Three-plus years at Toshiba, and the tooling that came out of it.
Software Engineer — HDD Division
Toshiba Software India Pvt. Ltd.
Current
Jan 2023 — Present
Bengaluru, India
- Develop and maintain stress-testing software that validates firmware features of SMR and Hybrid-SMR SATA hard drives.
- Contributed to a GUI-based HDD debugging tool that measurably improved firmware-debugging efficiency.
- Work hands-on with performance-benchmarking tools to analyse the key metrics that define a drive.
- Build scripts and tooling for parsing, visualising and analysing raw HDD performance data.
Selected tooling
HDD Stress-Testing Tool
A Linux stress-testing tool in C that validates SMR HDD firmware by hammering it with hours of I/O and zone-operation workloads — catching defects so releases ship stable and robust.
- SMR firmware validation
- I/O & zone workloads
- Long-duration soak tests
HDD Snapshot / Debug Tool
Windows GUI features built on the wxWidgets framework that let developers analyse firmware data extracted from HDD failure dumps — making the un-readable readable.
- Failure-dump analysis
- Native Windows GUI
- Faster firmware debug
HDD Performance Tool
Studied benchmarking tools like FIO and VDBench to evaluate drive performance, then built CLI and web tools to parse, visualise and analyse the data they produce.
- Benchmark evaluation
- CLI + web tooling
- Performance visualisation
05On-site · Silicon Valley
From Bengaluru to the Bay.
The work doesn't stay in one timezone. Keegan collaborates closely with Toshiba's US counterparts — and has been on the ground in California, from the Golden Gate to the Stanford quad — bridging an Indian engineering team, a Japanese multinational and a Silicon-Valley partner.
Golden Gate · San Francisco
A genuinely global engineer
One foot in three countries — an Indian engineering base, a Japanese multinational, and a Silicon-Valley partner. It's a workflow built for collaboration across timezones, languages and cultures.
India
Based in Bengaluru — engineering home base.
USA
On-site collaboration in California · Silicon Valley.
Japan
Japanese multinational employer · JLPT N5 certified.
06Research & AI
Beyond firmware — machine learning.
Peer-reviewed deep-learning research, and a genuine, hands-on interest in where modern AI is heading.
Published · Springer Nature
DOI 10.1007/978-3-031-68905-5_10
Development of a Pothole Detection System Using Deep Learning Techniques and Depth Estimation
Deep Sciences for Computing and Communications · Communications in Computer and Information Science (CCIS, vol. 2176)
Peer-reviewed research that uses deep learning to detect potholes and estimate their depth from imagery — a real-world computer-vision system, assessed on a purpose-built preprocessed dataset.
pothole · 0.94
pothole · 0.87
crack · 0.71
Working knowledge
Core ML foundations
Supervised & unsupervised learning — classification, regression and clustering.
Neural networks & deep learning
Architectures, and how modern language models are trained and fine-tuned.
Large Language Models
Their capabilities, limitations and real-world applications across domains.
Retrieval-Augmented Generation
Grounding LLMs in external knowledge to improve response accuracy.
Generative AI & prompt engineering
Following the landscape — generative models and how to steer them well.
AI for software & automation
Applying AI tooling to solve real problems and streamline workflows.
07Recognition
Awards & education.
Honours & awards
The Hult Prize
Second Runner-Up
MozoHack 3.0
First Runner-Up
ThinkTech 2k21
Winners
Hack With Us
First Runner-Up
SRMIST Merit Scholarship
Performance-based award — ₹43,750
Education
BTech — Computer Science Engineering
2019 — 2023SRM Institute of Science and Technology
CGPA 9.54 · Specialisation in Software Engineering (Agile, Software Quality Assurance)
Chennai, India
Higher Secondary (HSC) — Science
2019Loyola Higher Secondary School
78.5%
Goa, India
SSLC — Class 10
2017Loyola High School
89%
Goa, India
08Contact
Let's build something reliable.
Hiring for storage, firmware, performance or applied ML — or just want to compare notes on where AI is going? The fastest way to reach me is a message.
- SSD & next-generation storage engineering
- Firmware validation & performance tooling
- Applied AI / machine-learning engineering
- Based in Bengaluru — open to relocation