À propos
Job Title: Senior Python Backend Engineer - LangChain Agent Workflows & Government Data Pipelines
We do not respond to Automated messages
About the Project
GovGPT, an AI-powered platform (Already Built) that modernizes how government contractors find opportunities, conduct market research, and write proposals. Think of it as the AI layer that sits between government procurement databases , FPDS, USAspending) and enterprise proposal workflows (SharePoint, Word, Excel).
Our users are government contractors who currently spend weeks manually searching government contracts, copying data into spreadsheets, and writing proposals in Word. We're automating this entire workflow with autonomous AI agents.
What We're Building
Intel - Autonomous Research Agent (Core Product)
An AI research assistant that conducts market intelligence for government contracting. This is not a simple chatbot - it's a multi-agent system using the ReACT (Reasoning + Acting) framework.
How Intel Works:
User asks: "What counter-drone contracts has the Air Force awarded in the last 12 months?"
Intel breaks this into sub-tasks: identify keywords → query API → filter by agency → check FPDS for award data → cross-reference with USAspending → synthesize results
Agent decides which tools to call in sequence based on query type
Routes between Claude, GPT-4, and Gemini based on task complexity
Returns structured answer with citations to original government documents
Technical Requirements:
Agent orchestration using LangChain and LangGraph
Multi-step reasoning - agent decides next action based on previous results
Tool cascade - sequential API calls with fallback logic (GovTribe → → FPDS → USAspending → web search)
Multi-model routing - intelligently route queries to Claude/GPT/Gemini based on query type
Stateful conversations - remember context across message sessions
Citation system - track which API/document provided each piece of information
PDF integration - extract text, highlight cited passages, open documents inline
Data Hub - Document Library with Semantic Search
Government solicitations come with hundreds of pages of PDFs. Users need to find relevant past proposals, capability statements, and technical documents.
What You'll Build:
Document ingestion pipeline - process PDFs, Word docs, Excel spreadsheets
Metadata extraction - auto-tag documents by agency, NAICS code, contract type, dollar value
Semantic search - not keyword matching, actual semantic understanding using embeddings
Vector database implementation - Pinecone, FAISS, or Weaviate for fast retrieval
Tagging system - like GovDash (competitor), users can organize docs with custom tags
RAG pipeline - chunk documents, generate embeddings, implement hybrid search (semantic + keyword)
Document relationships - "Show me all proposals where we partnered with Lockheed"
Capture Search & Smart Alerts
Real-time monitoring of government procurement opportunities across federal, state, and local levels.
Data Pipeline Requirements:
API integrations: , , , GovTribe, 50 state/local procurement portals
Real-time ingestion using Apache Airflow or Kafka
Data transformation - normalize schema across different government sources
Incremental updates - detect new opportunities, track modifications to existing ones
Smart notifications - not just keyword alerts, semantic matching (user saves "cybersecurity," we alert on "network defense" contracts)
Deduplication - same contract appears on multiple portals, only show once
Proposal Generation Workflow
Intel doesn't just research - it helps write proposals by integrating with Microsoft Office.
Integration Requirements (This is Critical):
SharePoint integration - read/write proposal documents stored in client's SharePoint
Word document generation - create compliance matrices, proposal outlines, section drafts
Excel integration - extract pricing data, create cost models, populate templates
Claude + Office compatibility - leverage Claude's new native Excel/Word capabilities
Citation insertion - Intel researches a topic, writes a section, inserts footnotes with government source links
Version control - track changes, maintain proposal history
Collaboration - multiple users editing same document, Intel suggests content in real-time
Example Workflow:
User uploads solicitation PDF to Data Hub → Intel extracts requirements → generates compliance matrix in Excel → User asks Intel to draft technical approach section → Intel searches Data Hub for past performance examples → drafts section in Word → inserts citations → Saves to SharePoint, notifies team in Slack
Enterprise Integrations (For Major Defense Contractors)
We have a client (major defense tech company) who needs GovGPT integrated into their enterprise stack.
OAuth2 API Integrations Needed:
GovWin IQ - market intelligence platform (OAuth2, client provides credentials)
Microsoft SharePoint - document management (Microsoft Graph API)
Box - cloud storage (OAuth2 REST API)
Salesforce - CRM integration (OAuth2 REST API)
Slack - team notifications (OAuth2 REST API)
ProPricer - cost estimating software (Excel export/import or API if available)
Security Requirements:
NIST compliance - we handle CUI (Controlled Unclassified Information)
Tenant isolation - each company's data is completely separated
Audit logging - track every data access for compliance
Encryption - AES-256 at rest, TLS 1.3 in transit
OAuth2/JWT/RBAC - enterprise-grade authentication and authorization
Chatbot - (Legacy Feature - Being Replaced by Intel)
Currently we have a basic chatbot that answers questions about specific contracts. You'll be migrating this to the new Intel architecture.
Migration Requirements:
Move from simple Q&A to agentic workflow
Add tool-calling capabilities
Implement proper citation tracking
Connect to all data sources (currently only uses basic search)
Technical Stack (You'll Be Working With)
Backend & APIs:
Python primary language
FastAPI - REST API framework (we're already using this)
Flask/Django - acceptable alternatives if you prefer
GraphQL - nice to have for complex queries
gRPC - for internal microservice communication
AI & LLM Frameworks:
LangChain - agent orchestration (required)
LangGraph - multi-agent workflows (required)
OpenAI API - GPT-4, GPT-4o
Anthropic API - Claude Opus, Claude Sonnet (our primary model)
Google Gemini API - Gemini Pro
Hugging Face Transformers - for embeddings and custom models
Data & Databases:
PostgreSQL - primary relational database (via Supabase)
MongoDB Atlas - document storage for unstructured data
Redis - caching and session management
Vector Databases - Pinecone, FAISS, Weaviate, or Qdrant (your choice)
Elasticsearch/OpenSearch - full-text search (optional)
Data Pipelines & ETL:
Apache Airflow - workflow orchestration (preferred)
Apache Kafka - real-time streaming (alternative)
Apache Spark - batch processing (nice to have)
Pandas, NumPy - data transformation
Snowflake - data warehousing (future consideration)
Cloud & Infrastructure:
AWS - Lambda, ECS, S3, RDS (our current setup on Render)
Vercel - frontend hosting
Docker - containerization
Kubernetes - orchestration (if needed for scale)
Terraform - infrastructure as code
Security & Compliance:
OAuth2/JWT - authentication
RBAC - role-based access control
NIST - security framework implementation
Row-Level Security - Supabase RLS policies
Audit logging - compliance tracking
What We Need:
Must: Intel Agent Core
Set up LangChain/LangGraph agent architecture
Implement ReACT framework with tool calling
Build tool cascade logic (API fallbacks)
Create multi-model routing system (Claude/GPT/Gemini)
Test agent decision-making with complex queries
Must: Data Pipelines
Integrate , FPDS, USAspending, GovTribe APIs
Set up Apache Airflow for scheduled ingestion
Implement data normalization and deduplication
Build incremental update logic
Create monitoring dashboards
Must: RAG & Document Search
Design vector database schema
Build document ingestion pipeline
Implement chunking and embedding generation
Create hybrid search (semantic + keyword)
Add metadata tagging system
Must: Microsoft Office Integrations
OAuth2 authentication with Microsoft Graph
SharePoint read/write operations
Word document generation (compliance matrices, outlines)
Excel integration (pricing models, data extraction)
Citation insertion in documents
(Potentially) Enterprise Connectors
Salesforce OAuth2 integration
Box API integration
Slack notifications
GovWin IQ connector
Security hardening (audit logs, encryption)
Ongoing:
Optimize agent response times
Tune RAG retrieval quality
Add new data sources as needed
Performance monitoring and debugging
Work with frontend team on API contracts
What Success Looks Like
Intel can answer complex multi-step questions about government contracts
Agent successfully routes queries to appropriate tools
Citations are tracked and displayed
Data pipelines running daily, ingesting thousands of new opportunities
RAG system retrieving relevant documents with high accuracy
Smart alerts working for semantic query matching
Full Microsoft Office integration working
Users can generate compliance matrices from solicitation PDFs
Intel can draft proposal sections with citations
All enterprise integrations (Salesforce, SharePoint, Box, Slack) operational
NIST security controls implemented
Platform processing thousands of queries/day with sub-2s response times
Required Skills & Experience
Must Have (Deal Breakers):
3+ years Python backend development
Experience building agent-based AI systems (LangChain, AutoGen, or similar)
FastAPI or Flask/Django for REST APIs
PostgreSQL and MongoDB (dual database experience)
Experience with LLM APIs (OpenAI, Anthropic, or Google)
Data pipeline experience (Airflow, Kafka, or ETL frameworks)
OAuth2 authentication implementation
Can work independently with minimal supervision
Comfortable using AI tools (Claude, ChatGPT) to accelerate development
Strongly Preferred:
RAG systems - built production retrieval-augmented generation
Vector databases - Pinecone, FAISS, Weaviate, ChromaDB
Multi-agent workflows - LangGraph, ReACT framework, tool orchestration
Apache Airflow - workflow orchestration at scale
Microsoft Graph API - SharePoint, Word, Excel integrations
Real-time data streaming - Kafka, Flink, or similar
Enterprise SaaS integrations - Salesforce, Box, Slack APIs
Security/compliance background - HIPAA, FedRamp, SOC 2, NIST, or similar
Nice to Have:
Government contracting domain knowledge
Natural language processing (NLP) experience
Graph databases (Neo4j) for relationship mapping
MLOps experience (model deployment, monitoring)
Kubernetes and containerization
Experience with government APIs , FPDS)
What We're NOT Looking For
Full-stack generalists - we need backend depth, not breadth
ML researchers - we're building production APIs, not training models
Data scientists - we need engineering, not analysis
Frontend developers - we have a separate team
.NET/Java developers - Python is non-negotiable
Junior developers - this requires senior-level autonomy
AGENCIES
Work Style & Expectations
Communication:
Daily async updates - what you shipped, what's blocking you
Weekly sync call minutes, architecture review
Slack/Discord - primary communication (respond within 4 hours during work hours)
Documentation - everything you build gets documented in Notion
Speed:
We ship fast. First version of Intel needs to be functional in 3 days
Iterate quickly, don't overthink. 80% solution shipped beats 100% solution planned
Use Claude/ChatGPT aggressively to accelerate - we expect you to code 2-3x faster by leveraging AI
Autonomy:
You own the backend architecture decisions
Choose your own libraries/frameworks within our stack
Debug your own issues - we're a small team, limited hand-holding
Flag blockers early, but try solving problems independently first
Quality:
Code needs to work, doesn't need to be perfect
Security is non-negotiable (especially with NIST )
Write tests for critical paths (auth, data pipelines, agent logic)
Comment complex logic, but prefer self-documenting code
Project Timeline & Commitment
Duration: 3-month contract initially, strong potential for full-time conversion
Hours: Flexible, but expect 30-40 hours/week
Start Date: Immediately (within 1 week of hire)
Location: Fully remote (timezone preference: US-based for overlap)
Milestones:
Week 2: Intel agent functional demo
Month 1: Data pipelines operational
Month 2: RAG + Office integrations complete
Month 3: Enterprise connectors + security hardening
How to Apply
What to Include in Your Proposal:
Relevant Experience (Required)
Describe your most relevant project involving:
Agent-based AI systems OR multi-step LLM workflows
Data pipeline development (API integrations, ETL)
RAG or semantic search implementation
Technical Approach (Required)
Answer this: "How would you architect a multi-agent system that decides whether to query an API or search a vector database based on the user's question?"
LangChain Experience (If Applicable)
Share a specific example of building agent workflows with LangChain or similar frameworks
OAuth2 Integration (If Applicable)
Describe your experience integrating enterprise APIs (Salesforce, Microsoft, etc.)
Availability
When can you start? How many hours/week can you commit?
AI Tool Usage
How do you currently use Claude, ChatGPT, or other AI tools in your development workflow? (We expect you to leverage these heavily)
What NOT to Include:
Generic cover letters
"I'm a fast learner" (we need proven experience)
Salary/rate expectations (we'll discuss after technical fit)
Irrelevant experience (we don't care about your Java projects)
Why Work With Us
Technical Challenge: This isn't CRUD. You're building autonomous agents, not API wrappers.
Real Users: Defense contractors with $100M+ annual revenue are paying for this. Your code impacts actual multi-million dollar proposals. This also includes a full time spot if done properly.
AI-First: We use Claude to accelerate development. If you use any other LLM that is fine. You'll code faster than you ever have.
Ownership: You own the backend. Make architectural decisions. Choose your tools.
Impact: Government contracting is a $750B/year industry stuck in the 1990s. We're modernizing it.
Growth: 1-month contract → full-time → potential equity as we scale.
Questions?
If you're uncertain whether you're qualified, ask yourself:
Have I built systems where an AI agent decides what to do next?
Have I integrated 5+ external APIs in a production system?
Have I worked with LLM APIs beyond basic chat completions?
Can I set up a data pipeline that runs daily without manual intervention?
Do I regularly use AI tools to write code faster?
If you answered yes to 4/5, you're qualified. Apply.
If you answered no to 4/5, this isn't the right project.
Apply now with your relevant experience and technical approach
Contract duration of 1 to 3 months. with 30 hours per week.
Mandatory skills: LangChain, FastAPI, OAuth, Machine Learning, Python, RESTful API, PostgreSQL, MongoDB, Apache Airflow, API Integration
Compétences linguistiques
- English
Avis aux utilisateurs
Cette offre provient d’une plateforme partenaire de TieTalent. Cliquez sur « Postuler maintenant » pour soumettre votre candidature directement sur leur site.