Machine Learning Engineer · author of UB CSE TR 2026-43 · UB CSE Demo Day presenter
Buffalo, United States
Work formatOfficeHybridRemoteEverything above is an outside read of where you've been. This is the one block you write yourself — the direction you're steering toward.
Evidence shows graduate‑level research in personalized handwriting generation (UB CSE TR 2026‑43) plus a 2025 UB CSE Research Assistant stint and prior ML engineering role at EVE Healthcare. This mix supports applied ML and computer‑vision/generative‑model roles.
Career trajectory
Read purely by the companies on record — the same shortcut a recruiter takes.
engineering trainee
UNCONFIRMED
senior engineer / project delivery
UNCONFIRMED
machine learning engineer
UNCONFIRMED
research assistant (ML)
UNCONFIRMED
Trajectory moves from engineering trainee/project delivery into applied ML roles and UB research, showing a shift toward computer vision/generative models.
↗ Ascending⚠Employment steps are primarily LinkedIn‑reported with no third‑party confirmations.
Documenting and sharing results from Open‑Set Personalized Handwriting Generation (UB CSE TR 2026‑43) and the HW_Gen demo.
I can help teams turn research ideas in computer vision into working ML systems — from defining evaluation to wiring up training/inference with experiment tracking. At UB I authored a technical report on personalized handwriting generation and presented the work at Demo Day, so I’m comfortable bridging prototypes and demos. I’ve also worked in applied ML settings, so I focus on reliable pipelines, metrics, and profiling to keep inference responsive and measurable.
›Designing MLflow‑tracked experiments end‑to‑end
›Profiling GPU inference to reduce latency
›Personalizing generative models for handwriting synthesis
Author of UB CSE Technical Report TR 2026‑43 on open‑set personalized handwriting generation building on DiffusionPen, advised by Prof. David Doermann. Presented HW_Gen at the UB CSE Demo Day (Davis Hall, May 5, 2026). Appears in UB SEAS Graduate Ceremony 1 (June 2026) among Master’s recipients, aligning with the 2024–2026 MS window. Earlier experience includes a 2025 Research Assistant stint at UB CSE and prior ML/engineering roles at EVE Healthcare and AdaniConneX per LinkedIn. Public footprint is early: no peer‑reviewed publications beyond the UB TR and no public repo tied to the UB email surfaced.
An outside read of the profile based on available evidence — what's strong, the one thing to tighten, and what else surfaced.
This reads as an early‑career ML engineer with a concrete UB research artifact (TR 2026‑43) and a departmental Demo Day showing, bracketed by a 2025 RA stint and short prior industry roles. The identity is cleanly anchored by a UB email in the TR and a commencement listing. Public footprint is light and validation rests mostly on university artifacts and a self‑reported LinkedIn timeline.
A named UB technical report with advisor credit plus a Demo Day presentation provide credible, time‑boxed evidence of focused work in handwriting generation.
Named UB technical report
TR 2026‑43 lists Atul Pandey with an advisor and UB email, tying identity to the work.
Public demo appearance
Demo Day card shows the project and author’s UB email at a dated, on‑campus venue.
Degree timing coherence
Commencement program timing matches the research window and RA stint.
Most role metrics are self‑reported and there’s no public repo tied to the UB email; external replication path is unclear.
CEDAR/CUBS lineage fit The project sits squarely in UB’s document/handwriting ecosystem (CEDAR/CUBS).
AdaniConneX entity confirmed The former employer’s JV name appears in an SEC proxy, supporting the entity label.
UB GitHub Enterprise context UB provides GitHub Enterprise, suggesting code may live behind SSO.
Web sources + resume claims · 5 roles need confirmation
University at Buffalo, Computer Science and Engineering — Research Assistant (part‑time)
0 of 1 · Ask →2025-01 — 2025-09
Per LinkedIn: modular PyTorch training/inference with MLflow tracking; reports 0.8141 F1 and ~20% GPU latency reduction across reviews.
EVE Healthcare — Machine Learning Engineer
0 of 1 · Ask →2023-08 — 2024-07
Per LinkedIn: PyTorch/MONAI services across modalities; Azure inference <800 ms with MLflow; validation reduced manual review 35% and improved conversions 22%.
AdaniConneX — Senior Engineer
0 of 1 · Ask →2021-12 — 2023-07
Per LinkedIn: built Python/SQL KPI pipelines for a data‑center deployment and automated Power BI reporting; cross‑team coordination.
AdaniConneX — Assistant Project Manager
0 of 1 · Ask →2021-06 — 2023-07
Per LinkedIn: supported project delivery and reporting for data‑center programs.
AdaniConneX — Graduate Engineering Trainee
0 of 1 · Ask →2021-06 — 2021-11
Per LinkedIn: early‑career rotation across engineering/project functions.
Departmental demo showcasing personalized handwriting generation results.
UB CSE technical report proposing open‑set personalization for handwriting generation building on DiffusionPen.
Presented HW_Gen at UB CSE Demo Day (Davis Hall, May 5, 2026); departmental showcase appearance.
Each read carries a compliment and a tension, with its proof right beside it. It gets sharper as you go.
University at Buffalo
MS Computer Science · Computer Science · 2024–2026
Advisor: Prof. David Doermann
Vellore Institute of Technology
B.Tech · Electrical and Electronics Engineering · 2017–2021
GOVERNMENT FILINGS (1)
PRESS & INTERVIEWS (2)
CORPORATE & PRODUCT (9)
VERIFICATION CHECKS — RAN CLEAN (1)
SOCIAL & DIRECTORIES (1)
Atul Pandey works as Research Assistant (part‑time) at University at Buffalo, Computer Science and Engineering, based in Buffalo, United States.
Atul Pandey is a machine learning engineer known for UB CSE TR 2026‑43 and the HW_Gen Demo Day showcase.
Documenting and sharing results from Open‑Set Personalized Handwriting Generation (UB CSE TR 2026‑43) and the HW_Gen demo.
Atul Pandey has worked at 3 companies on record:
Projects and ventures Atul Pandey is behind:
Skills listed on Atul Pandey's profile: machine learning engineer. I can help teams turn research ideas in computer vision into working ML systems — from defining evaluation to wiring up training/inference with experiment tracking. Can teach: Designing MLflow‑tracked experiments end‑to‑end; Profiling GPU inference to reduce latency; Personalizing generative models for handwriting synthesis.
Atul Pandey is open to ML research engineer roles, computer vision roles, collaborations on handwriting gen and demo invitations.
Atul Pandey's education:
Public mentions of Atul Pandey:
Atul Pandey's public sites and channels:
This profile is assembled from 14 public sources, last aggregated 2026-09-18. Every fact carries a label: Verified (backed by two or more independent sources), Vouched (confirmed by a person who worked with them) or Self-reported. 2 facts are confirmed by cited sources. It is a public-source professional profile, not a background check.
Controlled by Atul Pandey
A registered account is linked to this profile — a living profile a real person maintains, not a passive scrape.
