Senior Prototyping Engineer

Keka Technologies Private Limited; Bengaluru; On-site; INR 3,000,000 - 4,500,000; Full time

Job summary

Aivar is hiring a Senior Prototyping Engineer to rapidly convert customer problems into AI-powered prototypes. You will work with Forward Deployed Engineers and AI/ML teams to ship a working prototype in days, not weeks.

You will join a fast-paced, customer-facing role that requires strong full-stack skills, an AI-first mindset, and the ability to switch context across healthcare, fintech, and logistics domains without losing speed.

Qualifications

• 7–10+ years of full-stack development experience.
• Hands-on AWS experience (Lambda, EC2, S3, API Gateway, RDS, CloudWatch).
• Experience integrating LLMs/GenAI into apps.
• Experience with agentic frameworks (Lang Chain, Lang Graph, CrewAI, AutoGen).
• Ability to ship working prototypes under tight timelines.
• Ability to design clean APIs and quick integrations with external systems.
• Strong knowledge of real-time architectures (WebSockets/SSE).

Responsibilities

• Move customer use cases from whiteboard to working prototype in 3–5 days.
• Build lightweight full-stack apps showing AI-powered workflows, voice agents, or agentic decisions.
• Integrate LLMs, RAG pipelines and agent loops into MVPs.
• Collaborate with Forward Deployed Engineers and AI/ML engineers to turn problems into prototypes.

Skills

Full-stack development

AWS

LLMs integration

Agentic frameworks

API Design

WebSockets

Demo storytelling

Auth basics

Git version control

Domain speed-learning

Tools

LangChain

GraphQL

REST APIs

WebSockets

SSE

LangGraph

OpenAI / Bedrock / Claude

CloudWatch

Job description

Aivar is an AWS-backed AI-native company that ships enterprise agentic systems in weeks, not quarters. We build across three accelerator platforms - Convogent (Voice AI), Velogent (Agentic Workflow Automation), and Kubogent (AIOps) - for customers across healthcare, fintech, logistics, and insurance.

We don't consult on AI. We engineer it. Fast.

About the Role

We are hiring a Senior Prototyping Engineer to sit at the intersection of customer discovery and rapid delivery. You will be the person who walks out of a customer conversation and has a working prototype running before the next meeting.

This is not a "build it right the first time" role. This is a "build it fast, prove it works, hand it off cleanly" role. You will work directly with our Forward Deployed Engineers and AI/ML Engineers to turn ambiguous customer problems into working AI-powered prototypes - in days, not sprints.

If you have strong full-stack instincts, an AI-first mindset, and the ability to context-switch across healthcare, fintech, and logistics domains without losing speed - this role is built for you.

What You'll Do

Rapid Prototyping

• Take a customer use case from whiteboard to working prototype in 3-5 days using Aivar's accelerator platforms and AWS-native tooling
• Build lightweight full-stack applications that demonstrate AI-powered workflow automation, voice agents, or agentic decision-making - whatever the customer use case demands
• Integrate LLMs, RAG pipelines, and agentic frameworks into rapid MVPs that are demo-ready and technically credible
• Work alongside Forward Deployed Engineers who bring the domain context - your job is to make it real, fast

Delivery Handoff

• Write clean, well-documented code that the delivery team can extend - not throwaway hacks that collapse under production load
• Participate in architecture reviews to ensure prototypes are built on patterns that scale to production on Velogent or Convogent
• Identify and flag technical risks early - what will break when this hits real data volumes, real payer rules, or real enterprise infrastructure
• Build reusable components and patterns that accelerate future prototypes across customer accounts

What We're Looking For

Must Have

• 7-10+ years of full-stack development experience - you are comfortable owning the entire stack from API to UI when needed
• Hands-on experience building and deploying on AWS - Lambda, EC2, S3, API Gateway, RDS, CloudWatch at minimum
• Proven experience integrating LLMs or GenAI APIs (OpenAI, Claude, Bedrock, or equivalent) into working applications - not just demos
• Experience with agentic frameworks - Lang Chain, Lang Graph, CrewAI, AutoGen, or custom agent loops.
• Ability to build fast under ambiguity - you have delivered working prototypes in compressed timelines and can show the work
• Prompt Engineering & LLM Evaluation - Ability to write effective system prompts, implement chain-of-thought patterns, evaluate hallucination rates, and set up guardrails for production-grade LLM integrations.
• API Design & Rapid Integration (REST/GraphQL) - Ability to design clean APIs and integrate messy third-party/enterprise APIs quickly. Every prototype connects to external systems - this is core workflow.
• Real-time / Streaming Architectures - Hands-on experience with WebSockets, Server-Sent Events (SSE), and streaming LLM token responses. Required for Voice AI (Convogent) and agentic loop interactions.
• Demo & Technical Storytelling - Ability to present working prototypes to customers, narrate what the system is doing and why it matters, and handle live Q&A. This role is customer-facing - communication is non-negotiable.
• Authentication & Authorization Basics (OAuth2, API Keys, RBAC) - Enterprise customers expect auth even in prototypes. Must be able to wire up login flows, API key management, and role-based access - especially in healthcare and fintech contexts.
• Version Control Discipline (Git Branching, PR Hygiene) - Clean commit history, meaningful PRs, and branch strategies that enable smooth handoff to delivery teams. Prototypes that live in a single messy commit can't be extended.
• Domain Speed-Learning Ability - Demonstrated track record of rapidly acquiring domain context and shipping working software in industries you knew nothing about 2 weeks prior. The role demands context-switching across healthcare, fintech, logistics, and insurance without losing speed.

Strong Plus

• Experience with RAG architectures, vector databases (Pinecone, Weaviate, pgvector), and document intelligence pipelines
• Familiarity with workflow orchestration tools - Temporal, Airflow, or equivalent
• Prior exposure to regulated industry domains - healthcare, fintech, or insurance - where compliance and auditability matter in the design
• Experience with React or Next.js for lightweight front-end prototype interfaces
• Containerization and basic DevOps - Docker, GitHub Actions, basic CI/CD
• Multi-modal AI (Vision, Audio, Document Parsing) - Experience with image recognition, audio processing, or document intelligence models. Healthcare = medical images, logistics = shipping labels, insurance = claim documents - text-only LLM skills aren't enough.
• Agent Memory & State Management - Understanding of conversation memory architectures (short-term, long-term, episodic), tool state persistence, and context window management for multi-turn agentic systems.
• UI/UX Sensibility - Not a designer, but has taste. Can use component libraries (Shadcn, Tailwind, Material UI) effectively to build prototypes that look credible and convert customers.
• Observability & Structured Logging - Ability to add lightweight tracing and debugging (LangSmith, OpenTelemetry, CloudWatch Logs Insights) so prototypes can be diagnosed when they fail during live demos.
• Event-driven / Message Queue Patterns (SQS, EventBridge, Kafka) - Understanding of pub-sub, async task triggers, and event buses. Agentic workflows are inherently asynchronous - tasks trigger other tasks.
• Graph Databases / Knowledge Graphs (Neo4j, Neptune) - Familiarity with graph-based reasoning for complex enterprise use cases - entity relationships, compliance trails, and multi-hop retrieval beyond flat RAG.
• Serverless-first Architecture Thinking - Beyond basic Lambda usage - understanding when to go serverless vs. container, cold start implications, cost profiles, and Step Functions for orchestration.
• Technical Documentation & ADRs - Ability to write architecture decision records, data flow diagrams, and setup READMEs. Clean handoff is not just clean code - it's clean context transfer.
• Figma / Wireframing Literacy - Ability to read Figma files or whiteboard UI flows in customer conversations before building. Reduces rework and aligns expectations before code is written.

What You Will NOT Be Doing

• Writing production-grade enterprise software from day one - that is the delivery team's job
• Sitting in long planning cycles before writing a line of code
• Building generic AI demos that look impressive but don't solve a real workflow problem
• Working on a single product for 12 months - every few weeks brings a new customer, a new domain, a new problem

The Ideal Profile in Plain Terms

You have been the person on the team when a customer says "I wonder if AI could do X," has a prototype ready to show them by Thursday. You build in public, iterate fast, and hand off clean. You are energized by the variety of problems, not overwhelmed by it. You care about making the prototype technically honest — not just visually impressive.

Our Stack (What You'll Work In)

Layer Tools

Languages Python, TypeScript/JavaScript

AI/LLM AWS Bedrock, OpenAI, Claude, LangChain, LangGraph

Cloud AWS (Lambda, EC2, S3, RDS, API Gateway, CloudWatch)

Frontend React, Next.js

Databases PostgreSQL, DynamoDB, Pinecone

DevOps Docker, GitHub Actions, Terraform

Platforms Velogent, Convogent, Kubogent

Why Aivar

• You will prototype for Fortune 500 healthcare systems, global fintechs, and fast-growing logistics companies - all in the same quarter
• AWS-backed with Bessemer Venture Partners investment - early-stage growth with enterprise customer access
• Direct exposure to agentic AI at production scale - not a side project, the core business
• Small, fast, senior team - no bureaucracy between your idea and shipping it.