Freelance AI Developers in MumbaiFreelance AI Developers in Mumbai
ProjectsPeople
Shezaan Ansari's avatar
Shezaan Ansari
proMumbai, India
AI Video Producer | Brand Designer
$25k+
Earned
19x
Hired
5.0
Rating
280
Followers
KITTL_EXPERT

expert

AGENT_EXPERT

Expert

+4
AI Video Producer | Brand Designer
Cover image for AI Assistant Development for Sonic
AI Assistant Development for Sonic
2
32
Cover image for Support AI Agent for Cartelle
Support AI Agent for Cartelle
2
22
Cover image for A hunter sparing the wolf
A hunter sparing the wolf that was meant to be his target 🏹🐺 Built this entire cinematic short using Higgsfield AI. Standard prompts usually give you flat camera motion and broken faces, so I pushed Higgsfield’s Cinema Studio controls, depth maps, and multi-shot consistency to land this level of realism. The Breakdown: Visuals: Higgsfield AI (photorealistic fur, atmospheric snow, dynamic lens lighting) Sound: Layered diegetic FX (iron clinking, snow crunch, breath) + dynamic cello score If you’re looking for high-end AI video production, pre-vis, or cinematic ad concepts—I'm open for contract work. Drop a DM or hit the hire button! 📩
2
2
256
Cover image for STORM HERD — 30 seconds,
STORM HERD — 30 seconds, one continuous take, no cuts. A herd of white horses stampedes across the slate rooftops of a nineteenth-century city while a thunderstorm tears the sky open behind them. The camera never breaks: it starts at hoof height on a rooftop ridge, lifts into the air as the herd charges past, drops into a chase behind them, threads between two galloping bodies, climbs above the roof valley, then dives and flies backwards ahead of the leaders to the final leap off the roof edge into open air. The whole film is graded to a single cold palette — cyan-steel monochrome, blacks crushed to zero, highlights blown to pure white. The only light source in the entire 30 seconds is the storm. Five lightning strikes punctuate the take, and at the 16-second mark one of them blows the frame completely white for two frames, turning the horses into black cut-outs before the image recovers through a bloom haze. That flash is the hardest part of the whole piece — a full-frame whiteout is exactly where a model wants to cut, and holding the take through it was the technical problem I built the prompt around. Made in CapCut Video Studio with Seedance 2.5. Full prompt and workflow below.
3
4
614
AI Developer(2)
Anurag Nagare's avatar
Anurag Nagare
Mumbai, India
I’m an AI & Machine Learning engineer with expertise in deve
I’m an AI & Machine Learning engineer with expertise in deve
Cover image for One Gate: Voice-controlled Gmail, Calendar,
One Gate: Voice-controlled Gmail, Calendar, and Drive that can't fire the wrong thing Every "AI agent" demo shows a voice command turning into a sent email like magic, but almost nobody shows the part that actually matters: what stops it from sending the wrong thing. I built proof: an agent that runs your Gmail, Calendar, Contacts, and Drive by voice reply to a thread, schedule a meeting, archive an email, find a file hands-free end to end, but never fires anything irreversible without you saying so. The honest hard part isn't getting an LLM to sound smart, it's this: a transcript goes to LLM against a strict JSON schema and comes back as an ordered plan, every step tagged reversible or not and exactly one thing in the whole system is allowed to check that flag. Finding a thread, drafting a reply, creating a calendar event: those just run. The result covers real ground without ever feeling like it's guessing: forward or reply to email with the original quoted underneath, archive/label/trash, resolve a name to a real address through your Contacts first and your mail history as fallback, schedule an event that creates quietly and only emails the invite after a second confirmation, answer "what did John say" or "what's on my calendar Thursday" grounded in content actually fetched from your account not hallucinated. No backend, no server anywhere in the loop
0
23
Cover image for Everyone's building AR filters and
Everyone's building AR filters and calling it "computer vision magic." Almost nobody's asking what's actually happening underneath — that most of these effects are just clever masking, not detection. Here's proof. I built an invisibility cloak that runs entirely in the browser, no green screen, no chroma key, no model training. https://github.com/AnuragNagare/Ghost-frame
4
2
270
Cover image for What your attention heatmap isn't
What your attention heatmap isn't telling you Everyone's staring at attention heatmaps and calling it "interpretability." Almost nobody's asking whether a single attention map actually tells you what the model used to make its decision. It doesn't. Not on its own. A raw attention map from layer 8 shows you what layer 8 attended to. It says nothing about how that signal got mixed, diluted, or overwritten by every layer before and after it. Attention rollout fixes this — and I built a walkthrough to show why it matters. Here's what makes it more than a "pretty heatmap" demo: Instead of visualizing one layer's attention, I traced how information actually flows through the full transformer stack. → Every layer's attention matrix is extracted, per head, per token → Multi-head attention is averaged, then combined with the residual connection (identity + attention) — this is the step most tutorials skip, and it's the one that actually matters → The combined matrices are matrix-multiplied layer by layer, rolling attention forward from input to output → The result: a single map showing genuine token-to-token influence across the entire network, not just one layer's snapshot The overlay shows you everything: → Per-layer attention vs. rolled-out attention, side by side → Token importance scores overlaid directly on the input text → A comparison view: which tokens raw attention says "matter" vs. which ones rollout says actually matter → Head-level breakdown so you can see which heads specialize vs. which are noise No black box. No "trust me, the model looked here." Just linear algebra, applied honestly across every layer instead of cherry-picking one. Built with PyTorch + HuggingFace Transformers + Matplotlib. Runs on any pretrained transformer, fully offline. ⚠️ Important: attention rollout is an approximation, not ground truth. It assumes attention is the primary information pathway, which ignores MLP layers and can still mislead for very deep models. Treat it as a debugging lens, not proof of causality.
0
107
Cover image for Everyone's racing to add biometrics
Everyone's racing to add biometrics to logins. Almost nobody's asking what happens when you can't — or shouldn't — touch the sensor. Shared kiosks, clinical settings, accessibility needs, hygiene-sensitive environments. Fingerprint readers and face unlock assume contact or a stored faceprint. Sometimes you want authentication that touches nothing and stores no biometric image of you at all. So I built GestureAuth — a contactless authentication system where your "password" is a sequence of hand gestures performed in front of a standard webcam.
1
80
AI Developer(10)
Bhavya Shah's avatar
Bhavya Shah
Mumbai, India
Tech solutions with a product-first approach.
Tech solutions with a product-first approach.
Cover image for RewriterPro
RewriterPro
0
43
Cover image for PhraseCraft: Revolutionize your web presence with AI-driven cont
PhraseCraft: Revolutionize your web presence with AI-driven cont
0
29
Cover image for MuchBeta: Connecting writers with tailored beta readers
MuchBeta: Connecting writers with tailored beta readers
0
19
Placeholder project card media
View more →
AI Developer(2)
Nipun Roy's avatar
Nipun Roy
proMumbai, India
AI & Web Products, SaaS & Mobile Apps | Product Designer
$10k+
Earned
1x
Hired
5.0
Rating
9
Followers
AI & Web Products, SaaS & Mobile Apps | Product Designer
Cover image for The next billion-dollar fintech product
The next billion-dollar fintech product won't look like another dashboard. It will feel like an operating system for your entire financial life. Personal finance. Business. Investments. Loans. Insurance. Taxes. One product. One experience. This is the concept I've been working on. 👇 #Fintech #ProductDesign #UIDesign #UX #SaaS #Startup #BuildInPublic #WebDesign #Figma #Design
2
92
Cover image for Market Pulse 📈 A modern sales
Market Pulse 📈 A modern sales and marketing dashboard built to give teams a clear view of performance without the clutter. The platform brings together campaign analytics, sales reporting, order management, customer growth, engagement metrics, conversion tracking, spending insights, and performance dashboards into a single workspace. Every screen is designed to make it easier to monitor KPIs, identify trends, and make data-driven decisions from one place. Loved exploring a bold visual direction with this one while keeping the experience clean, intuitive, and easy to navigate. #DashboardDesign #SaaS #UIDesign
2
1
149
Cover image for ProjectFlow A project management platform we
ProjectFlow A project management platform we recently designed and built. It brings together project planning, task management, meeting scheduling, team collaboration, calendars, analytics, and progress tracking into one workspace. The goal was to create an interface that feels clean and intuitive while still providing everything teams need to manage projects efficiently. Really happy with how this one turned out.
0
114
Cover image for VantaMail - A desktop email
VantaMail - A desktop email client with a premium experience! Designed and developed as a modern multi-account email client focused on speed, privacy, and elegant workflows. Clean UI, seamless account management, and productivity-first design.
1
1
118
AI Developer(7)
Jinal Joshi's avatar
Jinal Joshi
Mumbai, India
Custom AI Assistants & Automated Workflows for Business
5.0
Rating
5
Followers
Custom AI Assistants & Automated Workflows for Business
Cover image for QuickServe — AI Natural Language
QuickServe — AI Natural Language Business Portal & Automated Service Hub The Problem: Traditional business portals require navigating multi-level menus just to check account metrics or raise a support ticket. QuickServe replaces that complexity with conversational actions backed by secure backend execution. Key Deliverables & Architecture: Natural Language Data Access: Queries live balances and recent transaction records directly from SQL Server using structured OpenAI tool calls. Conversational Service Actions: Triggers backend service tickets and workflows straight from the chat interface. Fault-Tolerant Fallback Layer: Rule-based fallback logic ensures critical business queries remain functional even if third-party AI APIs experience downtime. Tech Stack: .NET Core, React, SQL Server, OpenAI API.
0
53
Cover image for Mittu — Multilingual WhatsApp AI
Mittu — Multilingual WhatsApp AI Agent for Local Businesses The Problem: Small retail shop owners and local businesses in India don't want another bloated dashboard or complex management software. They want to type a quick message in their everyday language and get business handled immediately—logging orders, generating invoices, and checking sales without friction. The Solution: I built Mittu, a conversational AI assistant on WhatsApp designed specifically around how Indian retailers actually communicate. Mittu operates directly inside WhatsApp, allowing store owners to manage daily operations entirely in Hindi and Hinglish. Key Features & Capabilities: Conversational Order Intake: Parses multi-item orders with customer names and pricing directly from natural, free-form text. Automated GST Invoicing: Calculates tax breakdowns (GST @ 18%) and generates structured invoice numbers (INV-xxxx) on demand. On-Demand Business Intelligence: Summarizes daily and weekly revenue, top-selling items, and transaction logs via simple voice/text prompts. Natural Hindi/Hinglish Processing: Built to understand mixed-language phrasing, colloquial terms, and localized business terminology. Architecture & Tech Stack: Core Agent Logic: Python LLM Engine: Groq (LLaMA 3.3) for low-latency reasoning and conversational routing Database & Memory: Supabase for customer order history and transaction tracking Messaging Layer: Twilio WhatsApp Business API Pitch / Call to Action: Looking to build custom AI agents that automate customer operations or bridge English-only AI tools into local languages? Let's connect and build it.
0
52
Cover image for CMS: Secure Payments. Smarter Operations. A
CMS: Secure Payments. Smarter Operations. A payment integration is not just about connecting a gateway. For this project, I designed and implemented a wallet-based payment workflow where users can recharge their wallet, use the balance for services, and have every transaction tracked with clear status and refund controls. The goal was simple: make payments reliable, traceable, and easy to manage for both users and administrators. Built with .NET Core, React, SQL Server, PayU and JWT Authentication. I enjoy taking a business process that feels fragmented or manual and turning it into a simple, practical digital solution.
0
82
Cover image for Chakran CMS — Turning Business
Chakran CMS — Turning Business Operations Into One Simple Workflow I designed and built this SaaS platform to bring applications, payments, wallet transactions, document handling, and reporting into one place. The focus wasn’t just on building features — it was on understanding the business workflow and turning it into a secure, easy-to-manage solution for both users and administrators. Built with: .NET Core, React, SQL Server, PayU & JWT Authentication. If you have a business process that feels complicated, manual, or disconnected, I can help turn it into a streamlined digital solution.
0
78
AI Developer(1)
Pranjal Shukla's avatar
Pranjal Shukla
Mumbai, India
AI Automation Engineer helping businesses run on autopilot
New to Contra
AI Automation Engineer helping businesses run on autopilot
Cover image for Client Onboarding Automation GoHighLevel →
Client Onboarding Automation GoHighLevel → ClickUp via n8n Problem: Client onboarding was fully manual — leads had to be checked in GoHighLevel, tasks created in ClickUp, and team members assigned by hand. This caused delays, missed steps, and inconsistent onboarding. Solution: Built an automated onboarding system using n8n that connects GoHighLevel and ClickUp. The workflow captures new leads, updates CRM data, creates tasks, assigns team members, and triggers follow-ups automatically triggering instantly the moment a new lead is captured. Result: Eliminated manual onboarding steps, reduced errors, and improved process efficiency for the team zero manual intervention required.
0
5
Cover image for Retell AI Voice Agent Real
Retell AI Voice Agent Real Estate Lead Qualifier Built an outbound AI voice agent for a high-ticket real estate coaching company using Retell AI Conversation Flow nodes. The agent qualifies leads through multi-branch logic, handles state-based hard gates deterministically, discloses AI identity honestly, and routes calls to human directors when needed. Deployed via Retell API using Node.js. Result: automated, compliant lead qualification at scale freeing human directors to focus only on qualified, sales-ready conversations.
0
9
Cover image for AI WhatsApp Lead Automation LeadPilot
AI WhatsApp Lead Automation LeadPilot for Real Estate Brokers Real estate brokers were losing leads to slow manual follow-up on WhatsApp. Built an AI-powered WhatsApp automation system (n8n + WhatsApp Business API) that handles lead conversations, qualifies buyers, and sends automated follow-ups 24/7. Result: faster response times, higher conversion, and leads engaged even outside business hours no manual chasing required.
0
12
Cover image for AI Voice Agent Lead Qualification
AI Voice Agent Lead Qualification & Appointment Booking Client was missing after-hours calls and losing leads to slow follow-up. Built a Retell AI voice agent integrated with n8n and GoHighLevel that answers calls 24/7, qualifies leads through a structured conversation flow, and books appointments directly into the calendar no staff involvement needed. Result: zero missed calls, faster lead response, and appointments booked automatically around the clock.
0
17
AI Developer(1)