Principal AI Software Engineer

About EnCharge AI

EnCharge AI is building the next-generation AI platform, delivering a 10x step-function improvement in compute energy efficiency and performance for AI inference workloads via novel in-memory-computing architecture.

The Opportunity

You will own the layer running agents against files, tools, documents with permissions, memory, unattended execution, and real outputs. This harness integrates open-source and bespoke code to support evolving models.

Key Responsibilities:

• Own end-to-end harness architecture: agent loop, safe execution, context management, knowledge base, memory, permissions, orchestration, observability.
• Build unique components with no open-source equivalent (e.g., session semantics, enforced permissions, human-editable file-based memory).
• Keep pace with models through adapters, prompt formats, tool-call schemas, and benchmarking.
• Ensure reliable tool use across varying model qualities via validation, repair, retries, and fallbacks.
• Develop agents, tools, MCP servers for internal/customer use cases with pre-shipment security reviews.
• Build evaluation harnesses including task suites, regression runs, cost/latency metrics per task.
• Define interfaces: Session API, CLI, GUI, and endpoints for existing tools.

Qualifications:

• 10+ years in backend systems or ML infrastructure engineering.
• Strong Python plus one systems language (Go, Rust, C++).
• Production experience shipping agent loops with tool use and multi-step workflows.
• Hands-on RAG, context management, memory for LLM applications.
• Sandboxing, isolation, and permission models expertise.
• Familiarity with open-weight models, quantization, and serving choices.

Nice to Have:

• Open-source agent harness contributions (e.g., SWE-bench, Terminal-Bench).
• MoE serving familiarity (expert placement, tensor parallelism).
• LLObservability pipelines.
• Desktop/GUI application experience.

Contact:
Email: muday_bhaskar@yahoo.com