Enterprise software delivery since 2009·Scrum leadership since 2011·Governed AI delivery since 2025
Santhosh Reddy
Senior Scrum Master & Agile Coach for AI-Enabled Engineering Teams
15+ years helping enterprise software teams deliver predictable outcomes through Agile coaching, quality engineering, technical leadership, and AI-assisted engineering workflows since 2025.
Not a ceremony facilitator: a delivery lead who recovers late programs, turns quality around with evidence, and puts governance around AI before it touches the release.
Four programs, four different problems, measured outcomes.
Recovery, quality transformation, cross-platform coordination, and delivery predictability — each with the evidence in view.
TELUSTelecommunications · Video Streaming · 2021–Present
Scaling Delivery and Quality Visibility Across a 5-Platform Streaming Ecosystem
No one had a single trustworthy picture of release readiness across five platforms.
5 platforms, 5 client development teams, and an 18-member QA organization coordinated into approximately 29 releases a year under one shared delivery view.
5platforms
5client development teams
18-memberQA organization
→
~29releases a year, one delivery view
Shared release visibility · same-sprint quality · governed AI assistance since 2025
Capability demonstratedEnterprise coordination: running delivery, quality visibility, and AI governance across five platforms without fragmenting ownership.
View full evidenceHide full evidence
Situation
A large-scale video streaming ecosystem shipping across Android Mobile, iOS/tvOS, Web, Smart TV, and Android TV, with 5 client development teams, an 18-member QA organization, and approximately 29 releases a year.
Constraint
QA operated in silos, development and QA tracked work in duplicate, and testing ran behind development, so no one had a single trustworthy picture of release readiness across platforms.
Intervention
Established executive-level Jira and Confluence dashboards covering release readiness, defect trends, and automation health
Led cross-functional workshops aligning Product, Development, QA, and Scrum Masters on same-sprint testing and integrated delivery
Partnered with QA Automation teams across E2E, regression, smoke, and sanity suites
Introduced AI-assisted workflows for defect documentation, story creation, and release communications, governed by human review and traceability guardrails, beginning in 2025
The trade-off
AI adoption was gated behind governance rather than rushed: human review and traceability came first, so quality and delivery standards held while manual reporting effort went down.
From Hundreds of Incidents Per Quarter to Single-Digit Levels in ~8 Sprint Cycles
Production incidents arrived by the hundreds each quarter, with no Definition of Ready or Done standards.
The combined work with Engineering and QA leadership helped reduce quarterly production incidents from hundreds to single-digit levels over approximately 8 sprint cycles through Definition of Ready and Done standards, same-sprint testing, and measured Agile maturity.
Hundredsof production incidents per quarter
Definition of Ready / Done
Same-sprint testing
Measured Agile maturity
Single digitsper quarter, in ~8 sprint cycles
Capability demonstratedQuality transformation: connecting Agile practice to a measured, sustained drop in production incidents.
View full evidenceHide full evidence
Situation
First Scrum Master hired at QHR Technologies, supporting Accuro EMR, an electronic medical record platform used by family practitioners and hospitals, starting with 2 development teams.
Constraint
Production incidents were arriving by the hundreds each quarter, there were no Definition of Ready or Definition of Done standards, and testing lagged behind development.
Intervention
Introduced Scrum ceremonies, Definition of Ready, Definition of Done, and structured backlog refinement across the initial teams
Partnered with Engineering and QA leadership to move testing into the same sprint as development
Tracked production incident trends and Agile maturity so improvement was measured, not assumed
The trade-off
Coaching expanded from 2 teams to the wider organization only after incident-trend data showed the practices working: adoption followed evidence rather than mandate.
quarterly production incidents
Hundreds to single digits
sprint cycles to turn quality around
~8
teams coached as results proved out
2 to org-wide
AT&TTelecommunications · eCommerce · 2018–2020
Containing Sprint Disruption and Restoring Delivery Predictability on an eCommerce Team
An 8-member team absorbed unplanned requests every sprint, and commitments kept slipping.
5 to 10 unplanned requests per sprint moved through a dedicated intake board, triage, and an explicit trade-off decision, protecting the sprint commitment and improving predictability.
5–10unplanned requests per sprint
Dedicated intake board
Triage
Explicit trade-off decision
Sprint commitment protected
Improved delivery predictability without walling off the business
Capability demonstratedPredictability under pressure: intake governance that protects commitments without walling off the team.
View full evidenceHide full evidence
Situation
An 8-member eCommerce development team was absorbing 5–10 unplanned requests every sprint. Commitments kept slipping, burndown data showed it, and testing coverage lagged behind development.
Constraint
Every incoming request was being accepted without triage, architecture was often unsettled at planning time, and the team had a single automation resource.
Intervention
Created a separate intake board with structured triage for out-of-scope requests
Made trade-offs explicit in standups: planned work and ad-hoc requests reviewed side by side, with deliberate decisions on what displaced what
Enforced Definition of Ready so architecture and design questions were settled before sprint planning
Strengthened code-review discipline and helped QA build automation within the same sprint
The trade-off
Unplanned work was made visible and negotiated instead of refused or silently absorbed, protecting sprint commitments while staying responsive to the business.
unplanned requests per sprint managed through structured intake
5–10
BoeingAviation · Electronic Flight Bag · 2017–2018
Delivering 12 Customer-Critical Features Across ~6 Sprint Cycles Without Adding Headcount
The Windows stream was approximately 6 months behind iOS, with airline clients waiting and no extra headcount.
From approximately 6 months behind iOS, usage-based prioritization and shared specialist capacity delivered 12 customer-critical features in approximately 6 sprint cycles.
~6 monthsbehind iOS
Usage-based prioritization
Shared specialist capacity
12features in ~6 sprint cycles
Recognized with a Boeing Certificate of Achievement
Capability demonstratedProgram recovery: converting a delayed platform stream into a sequenced, resourced, and delivered roadmap, recognized with a Boeing Certificate of Achievement.
View full evidenceHide full evidence
Situation
Boeing Aviator, an Electronic Flight Bag platform used by pilots and first officers, shipped on iOS and Windows. The Windows stream had fallen approximately 6 months behind iOS, and airline clients were waiting on critical features.
Constraint
No additional headcount was available, and the work spanned one onshore and one offshore team that had to move as a single delivery unit.
Intervention
Drafted a phased recovery roadmap prioritizing 12 features by observed client usage on iOS: high first, then medium, then low
Engaged project and senior program managers to borrow Architecture, UX, BA, and QA capacity from the iOS team instead of hiring
Ran daily coordination bridging the onshore and offshore teams to keep sprint delivery aligned
The trade-off
Recovery was sequenced by client usage rather than backlog age, and capacity was borrowed rather than added, trading roadmap breadth for speed on the features clients actually depended on.
customer-critical features delivered
12
sprint cycles from behind to delivered
~6
02Problems I take on
If this list sounds like your delivery organization, we should talk.
Situations I have been brought in to fix, each linked to its evidence.
A program is behind and nobody trusts the recovery plan
Sequence recovery by what customers actually use, negotiate shared capacity instead of waiting for headcount, and report progress against a phased roadmap leadership can verify.
Quality problems surface in production, not in the sprint
Move testing into the same sprint as development, set Definition of Ready and Done standards teams actually follow, and track incident trends so improvement is measured rather than claimed.
Cross-team releases have no shared readiness picture
Build one delivery view across teams and platforms: release readiness, defect trends, and automation health in dashboards that engineering and executives read the same way.
Give unplanned work a front door: structured intake, explicit trade-off conversations, and visible cost, so the team stays responsive without silently losing its sprint.
Align both on one plan: shared tracking instead of duplicate boards, same-sprint testing instead of handoffs, and automation health made visible to everyone who ships.
Enforce Definition of Ready with teeth: acceptance criteria challenged before planning, architecture questions settled up front, and stories that are honestly buildable or honestly not ready.
Leadership finds out about delivery risk when the date slips
Translate engineering reality into executive-level delivery signals: risk surfaced early, dependencies mapped, and readiness reported in terms leaders can act on.
Teams want AI in the workflow, but nothing governs it
Adopt AI where it demonstrably reduces manual effort, and wrap it in guardrails: human review before anything ships, traceability on every output, and quality standards that do not move.
Each with the evidence a hiring manager would ask for first.
Primary role fit
Senior Scrum Master
15+ years across enterprise programs at TELUS, Boeing, AT&T, and QHR. PSM II certified.
Agile Delivery Lead
Coordinates 5 platforms, 5 client teams, and ~29 annual releases in a streaming ecosystem at TELUS.
Agile Coach
First Scrum Master hired at QHR; coaching grew from 2 teams to organization-wide on measured results.
Quality Transformation Lead
Drove quarterly production incidents from hundreds to single digits in ~8 sprint cycles at QHR.
Additional leadership strengths
Technical Program Coordination
Coordinated multi-team program recovery at Boeing: 12 customer-critical features in ~6 sprints with no added headcount.
Governed AI-Assisted Delivery
Established AI governance and delivery guardrails at TELUS since 2025, and builds governed AI systems hands-on.
Core disciplines
Agile Coaching
Scrum Leadership
AI-Assisted Delivery
Quality Engineering
Delivery Excellence
Engineering Governance
Technical Program Delivery
Continuous Improvement
04Operating model
From customer problem to production feedback.
Quality gated in-sprint, AI assistance governed by human review. Open any stage to see it in practice.
01
Customer Problem
Understand the real user or business problem before jumping into implementation.
Customer-first scoping at TELUS
In practiceHide
Customer-first scoping at TELUS
At TELUS, delivery planning began from customer outcomes: acceptance criteria were written to reflect real user needs across five streaming platforms before implementation was scoped.
Convert the problem into scoped work, risks, acceptance criteria, and clear validation paths.
Boeing Aviator recovery roadmap
In practiceHide
Boeing Aviator recovery roadmap
Drafted a phased recovery roadmap with usage-prioritized sequencing for the Boeing Aviator program, delivering 12 customer-critical features in approximately 6 sprint cycles.
Use AI tools to accelerate implementation while keeping architecture, review, and decision ownership human-led.
TELUS AI-assisted delivery, since 2025
In practiceHide
TELUS AI-assisted delivery, since 2025
Beginning in 2025 at TELUS, introduced AI-assisted workflows for reporting and backlog administration, with human-review guardrails and traceability across five product platforms.
Validate through automation, exploratory review, release checks, and production-monitoring feedback.
TELUS QA coordination and release readiness
In practiceHide
TELUS QA coordination and release readiness
Coordinated E2E, regression, smoke, and sanity automation across an 18-member QA organization at TELUS, driving same-sprint testing adoption and release readiness reporting.
Inspect outcomes, refine practices, and carry learning forward: closing the loop from delivery back to discovery.
Structured intake improvement at AT&T
In practiceHide
Structured intake improvement at AT&T
At AT&T, introduced structured intake practices that managed 5 to 10 unplanned requests per sprint, strengthening Definition of Ready discipline and reducing delivery disruption.
AI in the delivery workflow, with the guardrails stated up front.
Since 2025: AI-assisted delivery at enterprise scale, and working AI systems built hands-on. The interesting part is the governance.
In the enterprise
At TELUS, introduced AI-assisted workflows for defect documentation, backlog authoring, sprint and release reporting, and stakeholder communications using Claude Code and generative AI, reducing manual administrative effort across a five-platform delivery organization. Adoption came with governance: delivery guardrails defined before rollout, not after.
How assisted work ships
AI-assisted work starts from a defined input, passes deterministic rules-based validation and human review, stays traceable to its source and reviewer, and ships under an accountable owner.
Input
AI-assisted step
Deterministic validation
Human review
Traceability
Accountable owner
Human review
AI-assisted output is reviewed and approved by a person before it is published or actioned.
Traceability
Every assisted artifact stays traceable to its source and its reviewer. Nothing ships unattributed.
Unchanged quality bar
Delivery and quality standards hold. AI reduces manual effort; it does not lower the gate.
Built hands-on
Personal engineering projects, started in 2026, that apply the same delivery discipline to AI systems: validation gates, bounded failure handling, and operational visibility, in working code.
Personal project · 2026
Animation Engine
A deterministic 2D character animation engine producing YouTube Shorts from JSON scene definitions. An AI story compiler turns narrative prompts into fully animated scenes via Claude and OpenAI APIs, with forced tool-use and a deterministic validation-and-retry gate between the model and the renderer.
passing tests guarding the pipeline
1,078+
pull requests merged under review discipline
74+
live end-to-end AI compile
47 s
Why it matters hereThe same discipline applied to enterprise delivery — validation gates, bounded failure handling, and traceability — implemented in working code rather than recommended in a slide deck.
Engineering & governance detailHide detail
What made it hard
LLM output is nondeterministic; rendering must not be. Every AI-generated scene had to arrive as valid, schema-conforming structure — never free text — and produce frame-perfect, reproducible output across multi-character scenes.
Governance decisions
Forced tool-use: the model must emit a structured scene definition against a schema — free-text output is never parsed
Gate A validation: every AI output is structurally validated before it can touch the renderer, with a bounded, deterministic retry loop on failure
Provider-agnostic adapters: the same compile interface runs Claude or OpenAI, so no single vendor owns the pipeline
1,078+ regression tests guard every AI-facing change before merge
A local-only Python monitoring dashboard for an AI workstation: hardware utilization, Ollama model status, and registered AI agent health in a single auto-refreshing page. Read-only, loopback-bound, no external data exposure.
Why it matters hereOperational visibility thinking — the single-pane-of-glass instinct behind enterprise delivery dashboards, applied hands-on to AI infrastructure.
Engineering & governance detailHide detail
What made it hard
Aggregating heterogeneous health signals — process state, GPU telemetry, Ollama REST APIs, filesystem checks — into one stable, auto-refreshing view that stays trustworthy enough to act on.
Governance decisions
Read-only by design: the dashboard observes, it cannot mutate the systems it monitors
Loopback-bound by default, with LAN access opt-in behind password protection, session timeout, and brute-force lockout
Path-traversal protection on every file-serving endpoint
API contract tests and browser-level Playwright tests on monitoring endpoints
PythonFastAPIpsutilJinja2OllamaNVIDIA
This site is part of the evidence: it is built with AI-assisted engineering under the same rules — typed publication schemas gate what can ship, every metric carries a source reference, and human review precedes every merge.
06Professional journey
Fifteen-plus years of widening delivery scope.
From business analysis to Scrum leadership, Agile coaching, enterprise cross-team delivery, and since 2025, AI-enabled delivery governance.
I help engineering teams deliver software more predictably by combining Agile coaching, quality engineering, technical leadership, and practical AI adoption. My experience spans enterprise software delivery across telecommunications, aviation, healthcare, financial services, and eCommerce, with a focus on improving delivery confidence, engineering quality, and continuous improvement. Beginning in 2025, I expanded this work into AI-assisted engineering and delivery workflows, using Claude Code and GPT review patterns to improve implementation discipline, documentation, traceability, and maintainability.
Business analysis→
Scrum Master leadership→
Agile coaching→
Enterprise delivery & quality→
Governed AI delivery (2025)
Senior Scrum Master / Agile Delivery Lead
Enterprise delivery · governed AI since 2025
TELUS · 2021 – Present
Agile delivery lead for a large-scale video streaming ecosystem, coordinating cross-platform delivery, release readiness, quality visibility, stakeholder alignment, and governed AI-assisted delivery workflows.
Scrum Master / Agile Coach
Quality transformation
QHR Technologies · 2020 – 2022
First Scrum Master hired at QHR Technologies, initially coaching development teams and later expanding Agile and quality practices across the organization.
Scrum Master / Agile Coach
Agile coaching
AT&T · 2018 – 2020
Scrum Master and Agile Coach for an eCommerce product team, strengthening intake, planning discipline, and collaboration across development, QA, and product stakeholders.
Scrum Master
Delivery recovery
Boeing · 2017 – 2018
Scrum Master for the Boeing Aviator Electronic Flight Bag platform, leading recovery coordination across iOS and Windows delivery while aligning stakeholders around customer-critical work.
Scrum Master
Agile transformation
Sammons Financial Group · 2016 – 2017
Supported an Agile transformation across multiple software development teams moving from Waterfall to Agile delivery, improving stakeholder transparency, release planning, backlog management, and team performance visibility.
M.S. Information Technology, full-time study
Graduate study
Missouri Western State University · 2014 – 2016
Full-time graduate study in Information Technology, completed 2016.
Scrum Master
Scrum Master leadership
Avaya · 2011 – 2014
Facilitated Scrum adoption across multiple teams in a communications technology organization, with a focus on dependency management, release coordination, stakeholder communication, and Agile coaching.
Business Analyst
Business & delivery foundation
Flipkart · 2009 – 2011
Requirements gathering, stakeholder workshops, process and gap analysis, and UAT coordination for eCommerce initiatives.
07Credentials
Education and certifications.
Education
Master of Science (M.S.), Information Technology
Missouri Western State University, USA — 2016
Certification
Professional Scrum Master I (PSM I)
Scrum.org — 2017 · Lifetime
Certification
Professional Scrum Master II (PSM II)
Scrum.org — 2025 · Lifetime
08Contact
Start the conversation.
Open to conversations about delivery leadership, quality engineering, and AI-enabled team workflows.