Parth Patel — Machine Learning Engineer

I’m a Machine Learning Engineer focused on building intelligent systems that bridge cutting-edge research with real-world product impact. With a B.E. in Computer Science from BITS Pilani and an M.S. in Computer Science (Machine Learning specialization) from Georgia Tech, I combine strong theoretical foundations with hands-on experience shipping scalable ML systems. Over the past few years, I’ve worked across the full stack of document AI — from model research and experimentation to deployment in large-scale production pipelines. My work spans: 1. Document structure understanding (page segmentation, table structure recognition, layout modeling) 2. Multi-modal GenAI systems (Figure Q&A, attribution workflows, deep-research agents) 3. Model optimization for production (latency, COGs, CPU inference, distillation, quantization) 4. Confidence calibration & evaluation frameworks 5. Pipeline unification and system design across ML + services layers I enjoy solving problems that sit at the intersection of: 1. Machine Learning 2. System design 3. Product thinking 4. Optimization under constraints (latency, cost, scale) What I care about: 1. Shipping ML, not just training models. I focus on measurable improvements — whether that’s step-function quality gains in table structure recognition, enabling multimodal Q&A capabilities, or significantly reducing inference latency and compute costs. 2. Bridging research and product. I enjoy experimenting with foundation models (LLMs, VLMs, open-set detectors) and translating them into robust, production-ready systems. 3. End-to-end ownership. From data preparation and architecture search to model evaluation, service integration, deployment, and cross-team alignment — I like driving ideas through to impact. How I Work: 1. Data-driven and experiment-focused 2. Structured in problem breakdown and execution 3. Comfortable collaborating across ML, backend, product, and research teams 4. Focused on speed and quality 5. Genuinely curious about emerging AI systems and how they can create real product differentiation

Stackforce AI infers this person is a Machine Learning Engineer specializing in AI and document processing systems.

Location: Noida, Uttar Pradesh, India

Experience: 6 yrs

Skills

Document Ai
Machine Learning
Generative Ai

Career Highlights

Expert in bridging research and product impact.
Proven track record in shipping scalable ML systems.
Strong focus on measurable improvements in AI.

Work Experience

Adobe

Machine Learning Engineer 3 (2 yrs 4 mos)

Machine Learning Engineer 2 (2 yrs)

Machine Learning Engineer 1 (1 yr 4 mos)

Research Intern (6 mos)

IEEE

Reviewer (WACV 2021) (0 mo)

Birla Institute of Technology and Science, Pilani

Undergraduate Teaching Assistant (4 mos)

SAMSUNG R&D INSTITUTE INDIA - BANGALORE PRIVATE LIMITED

Summer Intern (2 mos)

Homi Bhabha Centre for Science Education (HBCSE), TIFR

Summer Intern (2 mos)

Education

Master of Science - MS at Georgia Institute of Technology

B.E.(Hons.) at Birla Institute of Technology and Science, Pilani

Parth Patel

Machine Learning Engineer

Noida, Uttar Pradesh, India6 yrs experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Expert in bridging research and product impact.
Proven track record in shipping scalable ML systems.
Strong focus on measurable improvements in AI.

Stackforce AI infers this person is a Machine Learning Engineer specializing in AI and document processing systems.

Contact

Skills

Core Skills

Document AiMachine LearningGenerative Ai

Other Skills

AI AgentsComputer VisionMachine Learning SystemsML Model DeploymentML Model DevelopmentMultimodal AIDeep LearningLarge Language Models (LLM)PyTorchCC++Python 3

About

Experience

6 yrs

Total Experience

3 yrs

Average Tenure

5 yrs 8 mos

Current Experience

Adobe

4 roles

Machine Learning Engineer 3

Promoted

Jan 2024 – Present · 2 yrs 4 mos

Led high-impact ML and GenAI initiatives across document AI and multimodal systems.
1. Improved table structure recognition quality significantly on scanned documents and expanded support to new object classes (e.g. formula).
2. Trained and optimized a compact CPU-friendly unified model achieving ~90%+ of prior quality with much lower latency and footprint.
3. Shipped multimodal Q&A support by integrating image understanding into GenAI workflows, improving answer quality while keeping latency and cost controlled.
4. Designed and implemented attribution workflows across text and figures for both non-scanned and scanned documents.
5. Contributed to early development of multi-step AI agents for structured document research.
6. Worked on a PoC - "Bring Your Own Template" feature in Acrobat's Generative Presentations workflow
7. Mentored interns and team members (in projects around on-device SLM training, font detection, etc.); presented work in internal summits.
8. Contributed to multiple patent filings across document AI and GenAI features.

Document AIAI AgentsComputer VisionMachine Learning SystemsML Model DeploymentML Model Development+5

Machine Learning Engineer 2

Promoted

Jan 2022 – Jan 2024 · 2 yrs

Worked on document structure understanding and model evaluation for large-scale PDF workflows.
1. Built and benchmarked object detection models for page segmentation and table structure recognition.
2. Designed objective evaluation metrics (mAP, F1, content-based metrics) and integrated them into training pipelines.
3. Developed confidence calibration techniques to improve reliability of model predictions.
4. Identified data inconsistencies in relationship modeling tasks and drove structured re-annotation efforts.
5. Filed multiple patents based on model calibration and document understanding work.
Also completed M.S. in CS (ML), Georgia Tech (4.0 GPA) alongside full-time role.

Machine Learning Engineer 1

Sep 2020 – Jan 2022 · 1 yr 4 mos

Part of Document Cloud AI Team. Worked on Form Field Detection in Acrobat Sign.

Research Intern

Jan 2020 – Jul 2020 · 6 mos · Noida Sector 132

Interned at Media and Data Science Research (MDSR) Lab, Adobe Systems, Noida under Balaji K (Principal Scientist).
Worked on self-supervised learning for training GANs (Generative Adversarial Networks) for improved latent space semantics and image quality and diversity.
Work accepted at WACV 2021 (Winter Conference on Applications of Computer Vision); Also submitted a patent for the same.
Technologies Used: PyTorch & TensorFlow.

Ieee

Reviewer (WACV 2021)

Oct 2020 – Oct 2020 · 0 mo

Reviewed 2 research papers (related to applications of GANs) as a part of second-round submissions.

Birla institute of technology and science, pilani

Undergraduate Teaching Assistant

Aug 2019 – Dec 2019 · 4 mos · Pilani, Rajasthan, India

Teaching Assistant for the course Neural Networks & Fuzzy Logic. Was involved in designing assignments for the course and mentoring students for the course project.

Samsung r&d institute india - bangalore private limited

Summer Intern

May 2019 – Jul 2019 · 2 mos

Worked on a Proof-Of-Concept for Multi-modal Bixby Interaction, having two stages: food image classification and subsequent Q&A on the predicted recipe.
Collected dataset of around 60,000 images of food items by scraping the website allrecipes.com. Fine-tuned MobileNetV2 for predicting recipe name given food image.
Developed heuristic-Q&A system to answer query on predicted recipe name, using textual info. about the recipe present in collected dataset
Technologies Used: BeautifulSoup, Selenium, Aiohttp, Keras, & NLTK.

Homi bhabha centre for science education (hbcse), tifr

Summer Intern

May 2018 – Jul 2018 · 2 mos · Mumbai Area, India

Increased the efficiency of Decentralized Distributable Disk of Offline Open Educational Resources, abbreviated as DOER (a cluster of servers packed in a hard disk). DOER targets rural and underprivileged areas (w/o internet connectivity) and provides them with easy access to free educational resources.
Implemented a smart caching service in the backend, reducing server response time by more than 50% and thus, ensuring better user experience on low end devices (as are common in target areas).
Technologies Used: Python & Django.