Ziverge - Agentic Coding Essentials for Engineers

AI is changing how software gets built, but real advantage does not come from speed alone. It comes from learning how to apply AI with structure, judgment, and engineering discipline.

Agentic Coding Essentials is an in-person training for engineers who want to move beyond informal AI-assisted coding and develop a more operational approach to working with coding agents. The program introduces a more deliberate model for increasing output while maintaining strong standards for review, verification, and human ownership.

Led by Ziverge, this training is designed for engineers who want to understand how AI can be integrated into real delivery workflows with greater control and accountability. It offers a practical introduction to the mindset and working model behind Agentic Coding, helping participants build a stronger foundation for using AI in serious engineering environments.

This is for engineers who want more than novelty or speed. It is for those who want to use AI in a way that is durable, disciplined, and genuinely useful in production-minded software delivery.

🌅 Day 1 AM - Supervisory Engineering Foundations

The engineer's operating model: delegate, review, own. Models, inference, context windows, tools, agents, and autoregression as operator mental models. Why fluency is not correctness. Ownership standard: if it ships, you own it. Tactical model/tool selection: heavier vs lighter models, IDE agents vs terminal agents, exploration vs constrained-edit tools. Data posture awareness before tool adoption.

🌅 Day 1 PM (early) - Context Engineering and Trust Boundaries

Context engineering as the core operator skill. Core moves: write, select, compress, isolate. Persistent instruction architecture: CLAUDE.md, AGENTS.md, repo-wide and path-specific instructions, task templates, review instructions. Trust boundaries: never trust retrieved text, never trust generated code, never expose secrets casually, never permit privileged actions without review. State management: handoff notes, durable artifacts, compact summaries, fresh-context restarts.

🌄 Day 1 PM (late) - Planning, Decomposition, and Task Design

Planning as an execution artifact. Decomposing backlog items into bounded, parallelizable, mergeable units. Interface-first work, dependency mapping, merge order, vertical slices. Validation planning immediately after decomposition. Delegation triage: ideal for agents, acceptable with constraints, must remain human-led. Task shaping: explicit objectives, non-goals, constraints, machine-checkable outputs, known verification paths.

🌅 Day 2 AM (early) - Operating Environment and Safety Harness

Git worktrees, tmux, session architecture, lane isolation, persistent working layouts. Harness design: tests, linters, type checks, smoke checks, repro scripts, secret scanning, review checklists. Output validation before piping into shells, configs, CI, or deployment. Lightweight supply-chain discipline: inspect new dependencies, prefer approved libraries, verify transitive risk.

🌄 Day 2 AM (late) - Parallel Execution, Review, and Reconvergence

Running multiple bounded lanes. WIP control, rotation discipline, state visibility, stop-the-line triggers. Review craft: hallucinated APIs, confident wrongness, stale assumptions, fake completeness, thin tests, accidental blast radius, ornamental abstraction. Reconvergence: merge sequencing, shared-contract stabilization, lane summaries, merge rationale. Reproducible merge gates: rerun checks after each merge, require clean-session reproducibility.

🏆 Day 2 PM - Capstone Exercise

Hands-on integration exercise: triage realistic tasks, shape bounded lanes, launch parallel agent work, review generated output, reconverge, and validate. Not scored — designed to surface each participant's gaps and give concrete direction for continued practice.

This training has been scheduled in San Jose to align with AI & Big Data Expo North America, creating an opportunity for attendees to extend the value of an already valuable trip.
While the conference offers broad exposure to enterprise AI strategy, implementation, and emerging directions, Ziverge’s training provides a more focused environment for engineers who want a practical introduction to operationalizing AI in real delivery settings. If you are already attending the conference, this is a strong way to add depth, applicability, and technical substance to your time on site.

Helping Companies Work, Build, and Scale in an Agentic Era
Ziverge helps companies increase output by combining modern AI capabilities with strong engineering judgment, practical execution, and intelligent operational design.

What began as a high-end software engineering and consulting firm has evolved into a broader AI-enabled company, helping clients apply these technologies in ways that create real business value.

Today, Ziverge works across three core areas: designing and building bespoke Automation & AI Solutions, embedding engineers trained in Agentic Coding into client teams, and training organizations to adopt Agentic Coding effectively.

This approach helps companies ship more, upskill the teams they already trust, and transform repetitive operations into intelligent systems built for real-world use.

Ziverge serves clients across industries and geographies, with a focus on technical excellence, operational leverage, and practical innovation.

*No Refunds Policy

• "If you want hardcore programming content, this is the place to be. The advent of LLM like ChatGPT is going to change what it means to be a programmer. Come to this to stay relevant." ~ Event Attendee
• "Thanks to all the organizing staff for working so hard. Packed with content, yet beginner friendly and welcoming." ~ Event Attendee

Event Venue & Nearby Stays

301 S Market St, 301 South Market Street, San Jose, United States

Concerts, fests, parties, meetups - all the happenings, one place.

Get App