Gaurav Mangal
Software Development Engineer
4+ years exp
PHP
C
Python
JavaScript
C++
UX & Visual Polish: 5/10
Responsive & Cross-browser: 4/10
Active 3 days ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
Frontend UI engineer (junior) focused on hand-written interactive web UIs and visually polished demos. The strongest proven skill is building interactive browser-based visualizations with concrete artifacts in sorting-visualizer (index.html, css/style.css and scripts/main.js/quick_sort.js) showing DOM-driven animation, design tokens and control flows. There is little evidence of modern component architecture, automated tests, server-state handling or advanced performance work in public code.
Technical skills
PHP
C
Python• Middle • 4y+
JavaScript• Junior • 4y+
C++
Python
FastAPI
C++
PyTorch C++
Frontend
Vite
React.js• 4y+
DevOps
Rest API
AWS
Docker
Vector
Design
Figma
AI/ML
AI Agents
Claude
Computer Vision
Fine-tuning
Transformers
Embeddings
LLM
PyTorch
BERT
Scikit-learn
Semantic Search
NLP
QA
Postman• 4y+
Selenium• 4y+
Databases
PostgreSQL
FAISS
Timeline
Graduate Assistant
•
Middle
North Dakota State University
•
Full-Time
Led an AI research effort in collaboration with the university’s coatings department to optimize universality of primers. Coordinated model design and experimental planning for the research work. Reviewed large volumes of student code submissions and supported students with debugging and clean-code practices.
Python
C++
North Dakota State University
Master's Degree •
Computer Science
• Developed autonomous AI-driven hyperparameter optimization framework for GPT-2, orchestrated by frontier LLMs (Claude).
• Claude Opus 4. 6 edited code, ran fixed budget experiments, evaluated validation bits-per-byte (val_bpb) and decided on keeping or discarding change (no human intervention), closed looping in (edit → train → evaluate → keep/discard) for 50+ experiments.
PyTorch
huggingface_hub
Transformers
DeepEval
WebGPU
Software Engineer
•
Middle
Incedo
•
Full-Time
Built an LLM-based semantic retrieval system to replace keyword matching for noisy, unstructured resume data. Developed scalable document cleaning and normalization pipelines to generate embedding-ready inputs. Improved LLM-assisted extraction and semantic ranking versus a keyword baseline, and served the solution through REST APIs with vector indexing and caching.
Python
PyTorch
BERT
FAISS
FastAPI
PostgreSQL
Vector
Embeddings
Semantic Search
LLM
Founder
•
Executive
Trishna
•
Full-Time
Designed an AI-powered marketing optimization platform focused on predicting campaign performance and recommending budget allocation. Built supervised ML workflows including feature engineering, cross-validation, and hyperparameter tuning. Implemented ad-copy generation, embedding-based keyword retrieval, and time-series forecasting exposed via FastAPI services and stored in PostgreSQL, along with Docker/AWS deployment support.
PyTorchsince 2024
LLMsince 2024
Embeddingssince 2024
FastAPIsince 2024
PostgreSQLsince 2024
Docker
AWS
Guru Gobind Singh Indraprastha University (GGSIPU)
Bachelor's Degree •
Computer Science and Engineering
• Published a research paper in International Journal (IJFMR) on Predictive Prefetching to shorten Web Page loading times.
• Developed techniques to predict web navigation patterns based on user behavioral analysis and preloading assets for optimization.
NLP
Statistics and Machine Learning Toolbox
Python
• Engineered an automated document understanding pipeline to extract and structure information from complex technical documents and containerized the application and integrated components to support a scalable document processing workflow.
• Built vector-based semantic search layer to connect entities, relationships, and document context for efficient discovery/retrieval.
JavaScript
Docker
LLM
EasyOCR
Computer Vision
Vector
Knowledge Distillation
Software Engineer Intern
•
Junior
Bonds India
•
Internship
Developed responsive front-end React components, including real-time search and filtering to improve user interactions. Automated unit and integration tests using Selenium and API validation to reduce release regressions. Improved QA coverage to a reported target level through test automation and integration checks.
Pythonsince 2022
Postman
Selenium
React.js
JavaScript
Government of India Hackathon by MoE – 2021
Mobile app
• Finalist out of 120,000+ teams.
• Presented a research project on leveraging ML models in front of Indian prime minister and a panel of industry experts.
• Innovated IoT devices with mobile applications to analyze the learning curves of children and develop improvement techniques.
RIOT OS
Swift
SwiftUI
PyTorch
Junior Frontend Developer
Confidence: Low UI Engineer
Frontend UI engineer (junior) focused on hand-written interactive web UIs and visually polished demos. The strongest proven skill is building interactive browser-based visualizations with concrete artifacts in sorting-visualizer (index.html, css/style.css and scripts/main.js/quick_sort.js) showing DOM-driven animation, design tokens and control flows. There is little evidence of modern component architecture, automated tests, server-state handling or advanced performance work in public code.
UI Component Architecture
2/10
How interface parts are built
Small, hand-rolled DOM components and imperative JS controls; no component model, composition patterns or reusable UI primitives beyond module-level functions.
Evidence
sorting-visualizer/scripts/main.js: runalgo / disable_buttons / generate_array - central DOM-driven controller
sorting-visualizer/index.html: algorithm buttons + controls structure (no componentization)
sorting-visualizer/css/style.css: visual tokens and component styles (author-defined design tokens)
Responsive & Cross-browser
4/10
Works on all screens and browsers
Reasonable responsive CSS and some cross-browser input styling; fluid layout via percent widths and a media query for narrow viewports, but no advanced responsive features (container queries / logical props) or i18n/RTL considerations.
Evidence
sorting-visualizer/css/style.css: @media (max-width: 860px) responsive layout rules and vendor range-thumb rules
sorting-visualizer/index.html: viewport meta, semantic control structure for accessibility and layout
Performance Optimization
2/10
Speed of the interface
No measured or deliberate performance engineering; simple DOM updates and CSS transitions used to keep animations smooth but no code-splitting, virtualization or bundle/metric evidence.
Evidence
sorting-visualizer/scripts/main.js: generate_array creates DOM nodes directly for each bar (no virtualization or batching)
sorting-visualizer/scripts/visualizations.js: div_update / vis_speed functions used for animated updates (simple animation strategy)
Accessibility & Semantics
3/10
Usable for everyone
Basic semantic markup and aria usage (labels, aria-label on nav and legend, alt text in info files); accessible patterns exist but no deliberate focus management, keyboard-handling for custom widgets or CI a11y checks.
Evidence
sorting-visualizer/index.html: nav aria-label, legend aria-label, label 'for' attributes on inputs and alt attributes on images
sorting-visualizer/scripts/main.js: controls disable/enable logic (shows some accessibility/interaction state awareness)
State Management & Data Flow
2/10
Managing data in the app
Very light client state: globals and imperative flows coordinate UI and algorithm scripts; no server-state discipline, cancellation, optimistic updates or formal state machines.
Evidence
sorting-visualizer/scripts/main.js: globals (divs, div_sizes, array_size) and runalgo/disable_buttons/enable_buttons flow
sorting-visualizer/scripts/quick_sort.js: algorithm functions operate on shared arrays and call div_update for visual state
UX & Visual Polish
5/10
Look and feel quality
Visually polished UI with design tokens, transitions, color system and control affordances; good UX for a small interactive demo but focused on single flows and not resilient edge states or undo patterns.
Evidence
sorting-visualizer/css/style.css: design tokens, gradients, transitions and well-styled controls (range, buttons, legend)
sorting-visualizer/index.html: controls (speed/size) and legend that reflect interaction states
sorting-visualizer/scripts/quick_sort.js: step-by-step color/state updates and final sorted coloring
Expertise
HTML & CSS• Junior
Industries
Education• Junior
Technologies
JavaScript• Junior • 4y+
Figma
React.js• 4y+
Vite
Recommendations
- Develop interactive educational visualizations and small teaching tools (sorting visualizers, algorithm demos, data-story widgets).
- Own polished, accessible UI pages and micro-interactions - refine keyboard/focus handling and ARIA on custom widgets.
- Harden single-page flows by adding explicit async-state discipline: request cancellation, race handling and simple state machines for complex interactions.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: