Anshul Jain

Machine Learning Engineer

Bengaluru, Karnataka, India7 yrs 9 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

8 years of experience in building AI systems
Contributed to US patent for ML-based formula detection
Led R&D on intelligent design systems at Adobe

Stackforce AI infers this person is a Machine Learning Engineer with a strong focus on Generative AI and Fintech solutions.

Contact

Skills

Core Skills

Generative AiComputer VisionDistributed ApplicationsMachine LearningNlpSoftware Development

Other Skills

PythonC++Transformer ModelsNatural Language ProcessingGenerative Adversarial Networks (GANs)PyTorchLarge Language Models (LLM)Deep LearningJavaAmazon Web Services (AWS)ElasticsearchApache KafkaKubernetesREST APIsAWS

About

Machine Learning Engineer with 8 years of experience building production-grade AI systems across generative AI, computer vision, and NLP. Currently at Adobe working on cutting-edge generative AI — from intelligent layout generation using diffusion models and VLMs on Adobe Illustrator, to AI-powered video and document understanding across Premiere Pro and Acrobat. Contributed to a US patent filing in 2025 for ML-based formula detection in documents. Previously a founding ML engineer at Rephrase.ai, where work on hyper-personalized AI video generation at scale contributed to the company's Series A raise before its acquisition by Adobe. Also bring hands-on experience building ML systems and distributed data pipelines at scale in the fintech domain, having worked across personal finance and digital banking products. Core expertise: Generative AI · Computer Vision · LLMs · End-to-end ML Systems · Distributed Architectures Always open to connecting with ML practitioners and researchers pushing the frontier of applied AI.

Experience

7 yrs 9 mos

Total Experience

1 yr 6 mos

Average Tenure

2 yrs 2 mos

Current Experience

Adobe

Machine Learning Engineer - 3

Mar 2024 – Present · 2 yrs 2 mos · India · Hybrid

Worked across three product teams at Adobe, driving ML innovation from research through production deployment.
── Adobe Illustrator (Current) ──
Leading R&D on an intelligent design reflow system that automatically adapts layouts across aspect ratios by spatially rearranging design elements — combining transformer-based diffusion models with off-the-shelf VLMs in an integrated generative pipeline.
── Digital Video & Audio / Premiere Pro ──
Pioneered prompt-based, object-aware video autoreframe — a net-new capability enabling users to specify reframe attention targets via natural language prompts.
Built an automated video quality evaluation framework using VLMs and SOTA saliency detection models, reducing manual evaluation effort by ~75% (from ~2 engineer-hours daily to exception-based review only).
── Document Cloud AI / Acrobat ──
Built and integrated an ML model for formula detection and placement in document structure extraction; US patent filed (2025).
Expanded scanned PDF support in the Acrobat pipeline, increasing document coverage to 90% with 50%+ reduction in processing latency.
Enhanced ML and heuristic-based tagging for PDF components, improving document structure extraction accuracy.
Designed a multi-agent proof-of-concept to evolve Acrobat AI Assistant beyond single-turn chatbot interactions.

PythonGenerative AIC++Transformer ModelsNatural Language ProcessingGenerative Adversarial Networks (GANs)+4

Greenlight

Software Engineer

Jul 2023 – Mar 2024 · 8 mos · Bengaluru, Karnataka, India · Hybrid

Contributed to banking platform modernization, leading the rearchitecture of the spend platform from monolith to microservices to improve system scalability and resilience.

Distributed Applications

Jupiter

Machine Learning Engineer 2

Oct 2021 – Jul 2023 · 1 yr 9 mos · Bengaluru, Karnataka, India

Built ML-powered personal finance products in Jupiter's data science team, working across the full stack from model development to production APIs.
Developed an end-to-end transaction processing pipeline handling 5M+ daily records in real time, with ML-based categorization achieving 85% accuracy using metadata embeddings.
Designed a real-time anomaly detection system across multiple behavioral dimensions, enabling proactive financial alerts for users.
Built the gateway service for the "Money Tab" feature — delivering net worth, spend summaries, and personalized insights — with Redis/Kafka-based TTL caching reducing API latency by 70%.
Developed a personalized nudge product with a streaming ingestion pipeline (AWS Lambda + SQS) served via REST APIs.
Led cloud infrastructure cost reduction initiatives including smart pod autoscaling, DB load optimization, and caching strategies — reducing operational costs significantly.

NLPJavaAmazon Web Services (AWS)Machine LearningElasticsearchDistributed Applications+5

Rephrase.ai (acquired by adobe)

ML Research Engineer

Feb 2020 – Sep 2021 · 1 yr 7 mos · Bengaluru Area, India

Joined as one of the founding ML engineers at Rephrase.ai, an early-stage AI video generation startup, in a 25-person team. The platform was later acquired by Adobe.
Researched and engineered hyper-real talking head generation from text and audio using GANs and deep learning.
Built a production pipeline for realistic lip-sync and head motion synthesis using CNNs and autoencoders.
Led design and development of a scalable video generation platform producing ~5,000 hyper-personalized videos daily, built on Python, Django, and a microservices architecture.
Compressed image data assets by up to 75% through video compression algorithm experimentation, significantly reducing storage infrastructure costs.
Built end-to-end ML training and inference pipelines on Kubernetes with PyTorch for multi-tenant model serving.
Collaborated directly with enterprise customers to tailor video generation solutions to specific use cases.

PythonNLPGenerative AIDistributed ApplicationsTransformer ModelsKubernetes+6

Vmware

Member Of Technical Staff

Jul 2018 – Feb 2020 · 1 yr 7 mos · Bengaluru, Karnataka, India · On-site

Contributed to design and development of a complete SDDC (Software-Defined Data Center) stack solution for private datacenters.
Built an asynchronous telemetry collection system in Java/Springboot for performance monitoring and feature usage analysis.
Developed PoCs for cross-region and cross-cluster management of multiple SDDC environments.
Developed an automated file-based backup and restore solution for enterprise environments.

PythonJavaSpring BootREST APIsSoftware Development

Samsung electronics

Network Engineering Intern

May 2017 – Jul 2017 · 2 mos · Bangalore, India · On-site

Contributed to the SATP (Self Automated Testing Platform) framework for LTE Release 13, adding support for eMTC (enhanced Machine Type Communication) feature testing for IoT-connected devices.