We are looking for a senior, enterprise-grade Software Engineer to build a high-performance Proof of Concept (POC) for an Autonomous B2B Insurance Compliance & Asset Monitoring Agent. The system will bridge the gap between complex commercial insurance policy guidelines (unstructured data via RAG) and real-time operational telemetry (structured IoT log streams in Snowflake). The goal is an intelligent agent that dynamically analyzes asset anomalies against policy definitions to flag compliance breaches instantly. This is a highly intensive 1-week, 50-hour sprint for an engineer who understands complex data schemas, time-series IoT data gravity, and deterministic agent routing. Core Architecture Requirements (What You Will Build) The Multi-Agent Orchestration Layer: Build an autonomous agent framework (using LangGraph, AutoGen, or native Python execution loops) capable of multi-step reasoning, tool-calling, and error self-correction. Snowflake & IoT Data Lake Integration: Grant the agent secure access to a Snowflake environment containing structured IoT asset monitoring data (e.g., machine telematics, temperature logs, usage hours, maintenance schedules). The agent must dynamically discover schemas and write optimized SQL queries to aggregate this data. Insurance Compliance Hybrid RAG Pipeline: Implement a semantic search architecture to parse and embed dense, unstructured insurance underwriting guidelines, warranty contracts, and B2B compliance policies into a vector store. Deterministic Tool Use & Evaluation: The agent must safely look up a specific policy compliance rule (via RAG) and then automatically generate and execute a targeted query against the Snowflake IoT data lake to verify if real-world asset telemetry violates that specific rule. Required Technical Stack & Expertise Languages & Core Frameworks: Python (FastAPI / Typer), LangChain / LangGraph, LlamaIndex. AI & Embeddings: OpenAI GPT-4o / Anthropic Claude 3.5 Sonnet APIs, vector databases (Pinecone, pgvector, or Snowflake Cortex search). Data Architecture: Deep production experience with Snowflake, cloud data lakes, time-series IoT data manipulation, and advanced text-to-SQL generation patterns. Observability: Implementation of tracing tools like LangSmith, Phoenix, or Arize to monitor prompt tokens, routing paths, and costs. Project Timeline & Deliverables (50 Hours) Phase 1 (Hours 1–15): Environment setup, ingestion of sample insurance policy PDFs into the vector store, and secure connection string setup to a Snowflake mock IoT dataset. Phase 2 (Hours 16–35): Core development of the multi-agent execution loop, prompt tuning for text-to-SQL generation against time-series data, and policy-to-telemetry mapping logic. Phase 3 (Hours 36–50): Observability integration, edge-case testing for hallucination mitigation (ensuring false compliance flags are eliminated), clean code delivery, and a walkthrough video demonstration. If this sprint goes smoothly, there is an immediate opportunity to extend this contract into long-term architecture maintenance, production scaling, and integration into our core platform. Please include real examples of multi-agent, IoT, or insurtech data pipelines you have personally built in your application.

I want to build a private multi-model RAG-based Opportunity Intelligence Agent. It should support document ingestion, opportunity-specific workspaces, vector search, source citations, multi-model routing across OpenAI, Claude, Perplexity, and possibly DeepSeek, and generate strategic recommendations from both uploaded files and live web research. This is intended to become a reusable base agent capable of knowledge retrieval, web research, multi-model orchestration, document analysis, citation generation, and agent clonding and configuration. It will be used for analyzing & strategy development for project opportunities, responding to RFPs, and proposal assistance, as well as other applications.

Type: Hourly, ongoing (part-time to full-time, room to grow) Stack you'll work in: Notion, Slack, HubSpot, Google Workspace/Gmail, Claude + other LLM APIs, Zapier/Make/n8n About us We're a fast-moving sports and fan-engagement startup. We're small, we ship quickly, and we want AI woven into how the whole company operates, not as a side experiment, but as the default way we work. You'd be the person who makes that real. What you'll do Map our current workflows across sales, marketing, ops, and content, then find the highest-leverage places to automate. Build automations and agent workflows that connect our tools (Notion, Slack, HubSpot, Gmail/Google Workspace) using platforms like Zapier, Make, or n8n plus LLM APIs. Design and ship AI agents for real jobs: lead routing and CRM enrichment, content drafting, customer/fan response triage, internal knowledge search, reporting digests. Stand up the connective tissue (prompts, integrations, guardrails, and monitoring) so automations are reliable, not brittle demos. Train and enable our team: build SOPs, run working sessions, and create lightweight docs so non-technical people actually adopt what you build. Help set our AI strategy and roadmap as we scale. You're a strong fit if you Have shipped real automations and AI agent workflows in production (not just prototypes). Are fluent with Zapier / Make / n8n and at least one major LLM API (Anthropic/Claude, OpenAI). Know your way around HubSpot, Notion, Slack, and Google Workspace integrations and APIs. Can write clean prompts and think in systems: edge cases, error handling, human-in-the-loop checkpoints. Can explain technical work to non-technical people and get them to adopt it. Communicate proactively and move fast without breaking trust on things that touch customers or revenue. Nice to have Experience taking a small company "AI-native" end to end. Background in sports and/or blockchain. Comfort with light scripting (Python/JS) when no-code hits its limits. How to apply In your proposal, please: Describe one AI agent or automation you built, the tools involved, and the measurable result. Tell us how you'd approach training a non-technical team to actually use what you build. This part matters as much as the build. Share your hourly rate and weekly availability. Proposals that skip these will be passed over. We're looking to start with a small paid task and grow the engagement from there.

Engagement Overview I am the CEO and principal attorney of a small law practice specializing in campaign finance, lobbying regulation, FARA, nonprofit law, and government ethics. My five-person team — a junior partner, two associates, and an executive assistant — recently integrated into a larger firm. I am looking for an experienced Claude/AI automation builder for a phased engagement to design, build, and deploy a suite of interconnected agents and automations. This brief covers three phases. Phase I (Inbox Triage) is the highest immediate priority and the natural starting point. Phases II and III follow sequentially. Strong candidates will be evaluated on Phase I but should demonstrate familiarity with the full roadmap. This is a paid engagement. Scope, timeline, and rate are open to discussion. Technology Stack Email: Gmail (personal Pro account — not firm infrastructure) AI: Claude (Anthropic) via MCP or API Task and project management: Notion (existing workspace; routing tables, matter tracking, and timesheet structure already in place) Calendar: Google Calendar Internal chat: Google Chat Document storage: Google Drive (primary); local hard drives on iMac and MacBook Pro (secondary) Matter management / DMS: iManage (larger firm system — integration via dedicated ingestion email address) Voice notes: Plaud (AI note-taker) Signing platform: TBD — candidates should ask during scoping Out of scope: Signal and iMessage — encrypted platforms with no API access; manual forwarding convention only Confidentiality Requirements This is a law practice. Attorney-client privilege and work product protection apply to all client communications and matter-related documents. These are not compliance checkboxes — they are professional obligations with real consequences. The successful candidate must: • Execute a non-disclosure agreement prior to engagement • Demonstrate genuine understanding of why data handling matters in a legal context — not just technically, but professionally • Never use client names, email content, routing data, or document content for training, testing, or demonstration purposes • Work exclusively within the client's authenticated accounts — no third-party data stores outside the approved stack • Design systems that minimize data exposure — process and route, do not store unnecessarily Generic proposals that do not address confidentiality specifically will not be considered.   Phase I — Inbox Triage Agent Real-time classification and routing of inbound Gmail, with a daily digest to the executive assistant. Objective The principal attorney's Gmail inbox receives high volumes of email across clients, matters, and categories of widely varying priority. The goal is an agent that processes every inbound message, classifies it, routes it to the correct person automatically, and ensures nothing drops — without overloading the executive assistant with triage work she should not be doing. Two-Stage Routing Logic Stage 1 — Sender Classification Every inbound email is classified against a tiered contact list maintained in a Notion database: MVC: Most Valuable Clients — 5 to 10 contacts. Highest priority. HVP: High Value People — 10 to 20 contacts. Some overlap with MVCs. Principal attorney, unless task-type rule applies All other clients: Roster managed in Notion with assigned attorney(s). Assigned attorney(s) per Notion client record Catch-All: Anyone not in the contact table — prospects, opposing counsel, vendors, bar association, etc. Generate executive assistant daily digest Stage 2 — Task-Type Classification (MVCs only) For MVC contacts, a second classification layer routes based on the nature of the request. Rules are client-specific. Examples: • Scheduling requests → Executive assistant • Contracts and approvals → Designated associate(s) per client record • Strategic and substantive legal matters → Principal attorney Task-type rules are defined per MVC client and must be configurable without developer involvement. Routing Table — Notion All contact and routing data lives in an existing Notion database. The agent reads from it at runtime. Required fields: • Contact name and/or email domain • Tier (MVC / HVP / Standard / Catch-All) • Assigned attorney(s) for Standard clients • Task-type override rules for MVCs The executive assistant must be able to add, edit, and re-tier contacts without touching code. This is a hard requirement. Routing Output Candidates should propose their recommended approach from among the following, based on current Gmail MCP capabilities: • Apply Gmail label and/or forward to assigned attorney's address • Create a pre-addressed draft for principal attorney review before sending • Log routing decision to Notion with email link and recommended assignee Please address this question directly in your proposal — it is a key evaluation criterion. Daily Executive Assistant Digest Once per day at a configurable time, the agent generates a digest delivered to a designated Notion page covering all catch-all emails from the prior 24 hours. Each entry includes: sender, subject, timestamp, and a one-line AI summary of the email's apparent purpose.   Phase II — 5 AM Daily Brief A structured morning brief delivered to Notion each day before 5 AM, aggregating schedule, tasks, workflow status, news, and forward-looking context. Objective The principal attorney starts each day across multiple locations and needs a single, consolidated view of what matters — professional and personal — without opening email. The brief is delivered to a dedicated Notion page and covers the sections below in the following order. Section 1 — Daily Schedule Full calendar for the day pulled from Google Calendar. All events, calls, and commitments in chronological order. Section 2 — Open Projects and Undone Tasks Two sub-sections: (a) MVC high-value work — open projects and incomplete tasks for Most Valuable Clients, filtered to substantive legal work only; and (b) Personal — all open personal projects and tasks without exception. Personal items are comprehensive by design: if it is not surfaced here, it will be forgotten. Source: Notion task and project database. Section 3 — Blocking What is the principal attorney specifically holding up? Items where others in the firm are waiting for a review, decision, approval, or action. Source: Notion matter and task records where assignee or status indicates the ball is in the principal attorney's court. Note to builder: this section requires careful logic design. The agent must infer from status fields and assignee data what is genuinely waiting on the principal attorney versus what is simply unresolved. Work with client during onboarding to define the exact field logic. Section 4 — News Digest Industry News Curated digest of overnight developments in: campaign finance law and FEC activity, election administration, lobbying regulation (federal and state), nonprofit political activity, and government ethics. Format: short summary of each item with a link to the full article. Aim for signal, not volume — 5 to 10 items maximum. US Political News 5 to 10 headlines with links covering: presidential politics, US Senate and House elections, and major gubernatorial races. Stories people are actually talking about, not wire service filler. Section 5 — Firm Workflow Matter-level status summary pulled from Notion, organized by client tier and activity: Status Definition Closed Completed yesterday Moving Action taken yesterday Paused No action yesterday Stuck No action in five or more days Client groupings: MVCs (non-high-value work), Standard clients (all work), and any other open matters. Section 6 — One Month Look Ahead Rolling 30-day forward view pulled from Google Calendar covering: regulatory filing dates and compliance deadlines, matter-level deadlines, client birthdays, holidays, and planned vacations or travel. Anything that requires preparation or awareness in the next 30 days. Section 7 — Personal Financial Summary (If Feasible) Summary of personal financial position pulled from Monarch Money, if an API or MCP connector is available. Candidates should investigate Monarch's API access and address feasibility in their proposal. If not currently feasible, this section is omitted without affecting the rest of the brief. Delivery Notion only — not email. A dedicated page refreshed each morning before 5 AM. Previous day's brief should be archived, not overwritten.   Phase III — Night Maintenance Three nightly agents that run after close of business: timesheet creation, document filing preparation, and Plaud note routing. All outputs are delivered to Notion for principal attorney review. Part 1 — Timesheet Creation Objective Each evening, the agent reviews the day's activity across three sources and populates a timesheet in an existing Notion template for the principal attorney's review and finalization. Sources • Google Calendar — all events and calls attended • Gmail sent items — emails sent that day, grouped by client/matter where inferable • Google Chat — internal messages sent, grouped by thread/matter where inferable Note to builder: Google Chat API access will need to be confirmed alongside Gmail and Calendar MCPs. Confirm availability and any OAuth scope requirements in your proposal. Output: Populated Notion timesheet using existing template structure. Principal attorney reviews each morning, adjusts entries as needed, and finalizes. The agent does not finalize — it drafts. Part 2 — Document Filing Objective Each evening, the agent surfaces documents created or edited that day for the principal attorney's review. The attorney flags finals, and the agent forwards them to the firm's iManage ingestion email address for filing. Sources • Google Drive — documents created or modified that day • Local hard drives — iMac and MacBook Pro Note to builder: local hard drive access requires a locally-running component (daemon, Claude Code instance, or folder-watching script) on each machine. Please address your proposed approach to this in your proposal. Alternative approach for consideration: a designated 'Ready to File' folder on each machine that syncs to Google Drive. The attorney drags filing-ready documents into this folder throughout the day; the agent watches the folder and processes from there. Simpler architecture, device-agnostic, and builds a consistent filing habit. Candidates should evaluate and recommend. Output: A Notion page listing all documents surfaced for that day, with document name, location, and last-modified time. Principal attorney marks finals. Agent forwards marked documents to the iManage ingestion email address. iManage filing is handled by firm IT from that point — no direct iManage API integration required. Part 3 — Plaud Note Routing and Archiving Objective: The principal attorney uses a Plaud AI note-taker on calls and meetings. Each evening, the agent pulls new Plaud summaries, routes them to the appropriate team members, archives a copy to Notion tagged to the relevant client matter, and deletes the underlying audio and transcript from Plaud's platform and the local device. Prerequisite — Plaud API Plaud API or webhook access is a prerequisite for this part. Candidates must investigate and confirm availability before scoping. If Plaud does not currently support programmatic access, this part will require a manual export step as a workaround — please address both scenarios in your proposal. Routing Logic: Similar in structure to Phase I inbox triage routing (MVC/HVP/Standard tiers with task-type overrides) but with distinct rules to be defined with the client during onboarding. Do not assume inbox triage rules apply directly. Archiving: One copy of each Plaud summary is saved to Notion as a note, tagged to the relevant client matter. Tagging logic to be defined during onboarding. Deletion: After successful routing and archiving, the agent deletes: (a) the audio and transcript from Plaud's platform via API, and (b) any local copies on the principal attorney's devices. Local deletion requires the same locally-running component described in Part 2. Candidates may propose a unified local agent that handles both Part 2 and Part 3 local operations.   What I'm Looking For Strong candidates will have: • Demonstrated experience building Claude-based automations or agents — not general AI experience • Hands-on experience with Gmail MCP, Google Calendar MCP, and Notion MCP (or equivalent API integrations) • Ability to build systems that non-technical users can maintain — editability and simplicity are as important as technical sophistication • Comfort with phased delivery — Phase I first, Phases II and III following sequentially based on performance • Experience with professional services clients (legal, financial, consulting) is a meaningful plus • Willingness to execute an NDA and work within a legally sensitive environment What to Include in Your Proposal Please address the following specifically. Proposals that do not engage with these questions will not be considered. • Your proposed technical architecture for Phase I — how you would connect Gmail, Claude, and Notion • Your answer to the Gmail MCP routing output question in Phase I (labeling vs. drafts vs. Notion logging) — what is actually supported and what do you recommend • Your assessment of Plaud API availability and your proposed approach for Phase III Part 3 • Your assessment of Monarch Money API feasibility for the Phase II financial summary section • Your proposed approach to local hard drive access for Phase III Parts 2 and 3 — daemon, sync folder, or other • A comparable project you have delivered — describe the client type, the stack, and what made it work • Your estimated timeline and rate for Phase I, and a rough order-of-magnitude estimate for Phases II and III • Confirmation that you are willing to execute an NDA prior to engagement I am looking for someone who has read this brief carefully and has a specific, informed point of view on how to build it. This is phase one of a longer automation roadmap and the right candidate will be a long-term partner, not a one-time contractor.

Overview We're building an open-source CLI gateway for multi-agent AI orchestration — model-agnostic, MCP-native, and designed to bring any agent framework online with a single command. The repo is active, well-documented, and growing. We need an engineer to accelerate integration coverage and help attract open-source contributors. The Work Build agent templates and runnable examples for LangGraph, CrewAI, and similar frameworks Add LLM provider support (Groq, Mistral, Gemini, etc.) to the Hermes runtime Write clean, contributor-friendly code that models good PR hygiene Submit work via fork → PR → merge workflow on GitHub You Are Strong Python developer with CLI tooling experience Familiar with at least one of: LangGraph, CrewAI, LiteLLM, LangChain Comfortable with open source GitHub workflows (fork, PR, issues, reviews) Self-directed — you read docs, ask good questions, and don't wait to be unblocked Nice to Have Experience with MCP (Model Context Protocol) Familiarity with SSE, OAuth 2.1, or agent credential management Prior open source contributions Engagement Part-time to start, 20 hrs/week Fixed milestones per integration delivered Potential to grow with the project To Apply Share your GitHub profile and one example of open source work or a project that shows your Python and agent framework experience. https://github.com/ax-platform/ax-gateway

We are seeking an expert in GoHighLevel to assist with setting up and optimizing our website chat bot AI agent. It is already connected to the website, we just need someone that can help us configure it properly so that potential clients can interact with it

I'm seeking an expert to develop an AI agent that will assist in managing my business. The AI should be able to handle tasks such as scheduling, data analysis, and customer communication. The ideal candidate will have experience in business management and AI development, and be able to deliver a functional prototype within a month.

We're looking for an AI & Data Analysis expert to lead the integration of intelligent tools within the business platform. You'll connect Google Ads, marketing data files, and operational data sources to build AI agents via Claude that support business decision-making across our pet retail operations. Key Responsibilities Design and configure Claude-powered agents using tool use, structured prompts, and automated workflows for data analysis Integrate the Google Ads API to extract campaign metrics and feed decision-making dashboards Ingest, clean, and structure CSV, Excel, and other marketing data formats for agent processing Generate automated narrative reports and actionable visualizations for the executive and marketing teams Maintain and iterate on data pipelines connecting advertising, sales, and inventory data Required Technical Skills Claude API / Anthropic MCP (Model Context Protocol) Prompt engineering and LLM tool use / function calling Google Ads API Python or JavaScript (for pipelines and integrations) SQL / PostgreSQL / Supabase Pandas / NumPy or equivalent data libraries REST API consumption and integration Advanced Excel / Google Sheets Nice to have: Google Analytics, BigQuery, Looker, Power BI Ideal Profile Proven experience building data pipelines or LLM-powered tools in a production environment Hands-on familiarity with the Anthropic API and agent/tool-use patterns Ability to translate raw data into clear, actionable business recommendations Self-directed — can propose and build solutions without exhaustive specs Initial Projects Campaign ROI Agent — connects Google Ads + business sales data to generate automatic performance alerts and recommendations Marketing File Pipeline — ingests CSV/XLSX marketing files and produces AI-generated summaries and insights Executive Dashboard — decision-support interface with Claude-generated action recommendations based on live data

**Overview** We are a fast-growing SaaS company with a lean engineering team (~10 devs) utilizing a modern Python (FastAPI/Django) and Node.js backend, React frontend, and PostgreSQL stack. We have already deployed an initial multi-model agent stack—Claude Code, LiteLLM gateway, Git worktrees, and MCP integrations. We need an expert to run an intensive architecture review and optimization session for our current infrastructure. We are not looking for someone to build a full-time, weeks-long project from scratch. Instead, we need a seasoned engineer who has shipped this exact type of infrastructure end-to-end to audit our setup, identify architectural gaps, and guide our team on hardened implementation. This project must move fast. If your timeline is measured in weeks, please do not apply. We want someone who looks at this scope, jumps into a review session, and delivers actionable architectural guidance in days. This starts as a focused, urgent consultation. However, we expect ongoing advisory work—follow-ups, architecture adjustments, and enhancement reviews—as the AI tooling landscape shifts. For the right engineer, this will turn into a recurring relationship. We are completely open to a fixed price per milestone or an hourly structure. **What You Need to Have Actually Shipped and Can Review (Not Just Read About)** * **Full Agentic Coding Harnesses:** The entire loop: orchestrator → subagent → CI gate → merge loops. * **Isolation Layers:** Configured execution layers (such as E2B, Modal, or secure Docker runtimes) as isolated sandboxes for AI-generated code. * **Parallel Claude Code Sessions:** Managed multiple simultaneous subagents on scoped tasks via Git worktrees. * **Self-Hosted LiteLLM Gateways:** Routing to multiple models (Claude, GPT, Gemini, DeepSeek). * **MCP Server Infrastructure:** Connected file system, PostgreSQL, Atlassian, and Slack tool layers for active agents. * **Agent Framework Structures:** Used CLAUDE.md, COMMON\_MISTAKES.md, subagent role definitions, hook scripts, and settings.json. * **Human-in-the-Loop Orchestration:** Built Plan Mode or equivalent approval gates before agent execution. * **Multi-Agent Frameworks:** 7-agent feature factory patterns or frameworks like LangGraph, CrewAI, or Autogen. * **Durable Workflow Engines:** Applied Temporal, n8n, or similar tools for long-running agent workflow execution. * **Mechanical Quality Gates:** Treating CI green as the ultimate gate for agent output quality. \[[1](https://manveerc.substack.com/p/ai-agent-sandboxing-guide)\] **Our Current Stack (What you are reviewing)** * **Backend:** Python (FastAPI / Django) & Node.js (TypeScript) * **Frontend:** React (Next.js) * **Database & ORM:** PostgreSQL / Prisma / SQLAlchemy * **Infrastructure:** Docker Compose, AWS (ECS/EKS) * **CI/CD:** GitHub Actions / GitLab CI * **AI Layer:** Claude Code with shared `.claude/` directory, CLAUDE.md, and LiteLLM gateway in Docker * **MCP:** Atlassian (Jira/Confluence), custom PostgreSQL MCP server, Slack * **Workflow Automation:** Temporal / n8n * **QA Automation:** Playwright / Autonoma **Scope of Work (Review & Advisory Only)** 1. **Comprehensive Audit:** Audit our current agent harness and identify architectural gaps against a production-grade standard. 2. **Sandbox Strategy Consultation:** Review our environment strategy to ensure highly secure, isolated execution runtimes for agent code runs. 3. **Workflow Hardening Review:** Evaluate our parallel agent workflow setup (Git worktrees, subagent role configs, hook scripts, and settings lockdown). 4. **CI Pipeline Integration Strategy:** Advise on wiring our sandbox execution layer into the existing CI pipeline so agent-executed code runs in clean snapshots, not live infra. 5. **Architectural Runbook:** Deliver an optimization report / documented standard that our backend lead can easily own and execute going forward. **How to Apply** Skip the generic pitch. Show us something real to be considered: 1. A GitHub repo, architecture diagram, or Loom walkthrough of an agentic harness you have actually shipped. 2. Specific tools from our stack you have personally configured (E2B, LiteLLM, Claude Code, etc.). 3. One sentence explaining the hardest problem you solved to get full agent loops running reliably. 4. Your availability to conduct this high-impact architectural review session this week.