LLM Observability System Implementation
Budget: $250 – $750 USD
Don't Apply before you have experience in Langfuse or similar tools
We rely on large language models (LLMs) to power several core product features. As we continue scaling, we’re investing in deeper LLM observability to gain insights into usage patterns, performance bottlenecks, and cost optimization opportunities.
Project Scope
We’re looking for a freelancer with hands-on experience instrumenting LLM observability tools in production-grade applications. You will work closely with our engineering team to design and implement a scalable, cost-aware observability system for LLM interactions.
Our Current Stack
Backend: Golang (monolith + services)
LLMs: Amazon Bedrock (Anthropic Claude) and OpenAI (GPT-4)
Hosting: Cloud-based API server
Frontend: Flutter (Web + Mobile)
What You’ll Do
Evaluate observability tools Langfuse for LLMs.
Propose an implementation roadmap based on Teachally’s goals and technical environment.
Instrument LLM calls across Bedrock and OpenAI to track latency, token usage, error rates, costs, user-level tracing, and prompt/response debugging.
Set up dashboards and alerting for real-time visibility into model usage and performance.
Ensure data privacy and efficiency, especially around logging user inputs and model responses.
Deliver well-documented code, configuration, and optionally provide handoff sessions for internal developers.
You Might Be a Good Fit If You Have:
Experience integrating LLM observability in production systems. Strong familiarity with tools like Helicone, LiteLLM, Langfuse, etc.
Working knowledge of Golang and experience instrumenting backend services. Comfort working with both OpenAI and Bedrock APIs.
Bonus: understanding of prompt evaluation, token usage monitoring, and fine-grained cost optimization.
To Apply
Please include:
A brief proposal outlining how you’d approach observability for our LLM systems. Examples or links to similar work you’ve done.
Availability or project-based rate.
We rely on large language models (LLMs) to power several core product features. As we continue scaling, we’re investing in deeper LLM observability to gain insights into usage patterns, performance bottlenecks, and cost optimization opportunities.
Project Scope
We’re looking for a freelancer with hands-on experience instrumenting LLM observability tools in production-grade applications. You will work closely with our engineering team to design and implement a scalable, cost-aware observability system for LLM interactions.
Our Current Stack
Backend: Golang (monolith + services)
LLMs: Amazon Bedrock (Anthropic Claude) and OpenAI (GPT-4)
Hosting: Cloud-based API server
Frontend: Flutter (Web + Mobile)
What You’ll Do
Evaluate observability tools Langfuse for LLMs.
Propose an implementation roadmap based on Teachally’s goals and technical environment.
Instrument LLM calls across Bedrock and OpenAI to track latency, token usage, error rates, costs, user-level tracing, and prompt/response debugging.
Set up dashboards and alerting for real-time visibility into model usage and performance.
Ensure data privacy and efficiency, especially around logging user inputs and model responses.
Deliver well-documented code, configuration, and optionally provide handoff sessions for internal developers.
You Might Be a Good Fit If You Have:
Experience integrating LLM observability in production systems. Strong familiarity with tools like Helicone, LiteLLM, Langfuse, etc.
Working knowledge of Golang and experience instrumenting backend services. Comfort working with both OpenAI and Bedrock APIs.
Bonus: understanding of prompt evaluation, token usage monitoring, and fine-grained cost optimization.
To Apply
Please include:
A brief proposal outlining how you’d approach observability for our LLM systems. Examples or links to similar work you’ve done.
Availability or project-based rate.