Akram Sheriff

I like to build and convert cutting-edge technologies into scalable, real-world products that solve hard business problems in Enterprise IT and other market adjacencies. I have worked for 17+ years at the intersection of AI, ML, Cloud, SaaS, Security, IoT, Enterprise Networking, Wireless, and Embedded M2M — not just as a technologist, but with a founder's mindset: identifying hard problems early, connecting dots across domains, and executing from first principles to deliver real outcomes.

I care about building 0→1 products for emerging market categories, AI-native systems, AI for Security + Security for AI, translating deep technology into product strategy and business impact, and pushing the innovation horizon through research and thought leadership. I like to approach Technology and Product Strategy by asking not only can we build this? but also should we, and for whom?

What I Build & Care About

• Building 0→1 products for emerging market categories
• AI-native systems: LLMs, agents, orchestration layers — scalable, observable, secure
• AI for Security + Security for AI
• Translating deep technology into product strategy and business impact
• Thought leadership, research, and pushing the innovation horizon
• Technology and Product Strategy — asking not just “can we build this?” but “should we, and for whom?”

Journey

My journey spans:

• 2005–09: B.Tech undergrad, Amrita University (Coimbatore) — University topper in final year and consistently in the top 1 percentile across all four years; Cisco Networking Academy Student Ambassador; completed CCNA and CCNP; part of Amrita's CEN (Center for Computational Engineering & Networking) Supercomputer Lab, working on parallel programming fundamentals (2008); ML + Wireless Networking + Signal Processing research and project work; internship at Robert Bosch under a visiting student exchange program in the EU region
• 2009–10: Embedded Systems, Wireless, IoT and Automotive software development
• 2011: Completed CCIE Wireless (Written) at Cisco Systems, Bangalore
• 2011–12: Embedded Systems, Wireless Networking, IoT Software Engineering, Routing & Switching
• 2012: RTLS and WiFi location tracking — led engineering of Aruba Networks' ALE (Analytics & Location Engine) startup product from scratch
• 2013: AoA + ML wireless / IoT positioning — extended ALE with Angle-of-Arrival and ML-based positioning techniques
• 2014–16: Cloud SDN-based Network Controller — Enterprise Networking — led launches of Aruba WiFi 4, 5 & 6 AP platforms and the Aruba Cloud SDN Controller, later integrated into HPE GreenLake following HPE's acquisition of Aruba Networks
• 2017: iPOWER — WiFi + BLE + ML IoT power-optimization SaaS platform; leveraged PSU/PDU telemetry and IoT sensor data (temperature, humidity, pressure) for continuous enterprise & hyperscaler datacenter monitoring, with a closed-loop sensor-activation approach for rack-level PSU/PDU power optimization (DCIM) — driving Opex cost savings and sustainable energy management; completed MIT IoT (Road to Connected Digital World) thought leadership immersion program
• 2018–19: Cloud-native AWS IoT platforms and TM Forum ODA AI digital twins
• 2019–20: IoT container management infrastructure for the Edge; boot-time optimizations with Linux containers
• 2020–21: Stanford Masters Research Project: ScaleAI for secured edge inferencing in MCUs, which led to OmniML — stealth-mode startup
• 2020–21: Leveraging Cisco SDA (Software-Defined Access) with a VXLAN fabric to bring an overlay architecture for wireless IoT access points, integrated with the Cisco DNAC controller for OT + IT convergence
• 2022–25: CDP, Data Lake, Data Infra in AWS Cloud, K8s security, CNAPP, CSPM, DSPM, CASB, Agentic AI Orchestration, AI Security, Agentic Deployment, Agentic Protocols for Collaboration, MCP Gateway Agentic Infra
• 2024: Researched, built, and led Cisco Outshift's first internal Agent Builder Enterprise Platform (ABP) project from scratch — delivering persistent, long-running, task-agnostic agents that are composable at runtime and then execute to achieve any business task or goal, usable by internal developers, IT admins, and non-technical users. Built the ABP evaluation framework, modeled on OTEL trace-based scoring with reasoning, action, and tool-call metrics to derive the OEC (Overall Evaluation Criterion)
• 2024–25: Building an Agent Builder Platform for the Enterprise at scale, serving different enterprise user personas; secured agentic collaboration & orchestration for the enterprise; built Agentic AI deployment infrastructure and Kubernetes-based runtime infrastructure projects at Cisco Systems
• 2025: Built an Agentic Context Graph for the Enterprise
• 2025–26: Customizing Voice Agentic Inference Infra engineering innovations and Infra product development — multi-cloud, multi-K8s-cluster, Neocloud infra with vLLM, SGLang, Envoy, and Service Mesh (Istio)-based infrastructure
• Recent: IoT, Kubernetes security, Agentic AI, MCP Gateway, MCP Infra, MCP Security, Agentic Protocol Infra, Multi-Agent Systems, GenAI security, Voice Agentic Inference Infra, Loop Engineering, Designing Agentic Harness, and multi-modal inference-time security

Current Focus

• Agentic AI Orchestration
• Agentic AI Security
• MAS / Multi-Agent Systems
• LLMOps
• ML Security
• LLM Model Security
• MCP Gateway Security
• Agentic Infrastructure — ANS, IIOA/IOC, Cisco AGNTCY, ANS + MIT NANDA
• Autonomous Agentic Observability
• Optimizing Agentic AI workflows during the inference stage based on traffic type
• Agentic commerce protocols — x402, ACP, UCP, AP2, MPP

Research, Open Source & Thought Leadership

80+ issued U.S. patents · 100+ filed U.S. patents · OSS contributor · AI Research · Conference speaker / Thought Leader

Awards & Recognition

• Cisco 50th Patent Award — 80+ issued U.S. patents (100+ filed) at Cisco Systems, presented "in appreciation of hard work and dedication to the advancement of technology." Recent grants include Data Security at Cloud Scale (No. 12,675,590, July 2026), Attack Detection and Mitigation for Fine Timing Measurement (No. 11,726,173, August 2023), Microservice Visibility and Control (No. 11,601,393, March 2023), Access Point Based Location System for High Density WiFi Deployments (No. 11,337,176, May 2022), and Traffic Class-Specific Congestion Signatures for Improving Traffic Shaping and Other Network Operations (No. 10,873,533, December 2020).
• Cisco IEC (Innovate Everywhere Challenge 2) — two New Product Innovation ventures entered in Cisco's Global Intrapreneurship Contest for employees: AIP (AI-enabled Switching Fabric Port Analyzer — neural-net-powered real-time classification / anomaly detection for core routing & switching products), Winner; iPOWER (WiFi + BLE + ML IoT power-optimization SaaS), Finalist (2017, San Jose).
• Cisco Pioneer Award for IoT
• Best Paper Award — IEEE 5th International Conference on AI in Cybersecurity (ICAIC-2026), University of Houston, February 2026, for Agent Name Service (ANS): A Universal Directory for Secure AI Agent Discovery and Interoperability (with Ken Huang, Vineeth Sai Narajala, and Idan Habler).
• 2nd Place, AI Safety & Alignment — AgentX Competition (Research Track), UC Berkeley RDI Agentic AI Summit 2025 (with Ken Huang, Vineeth Sai Narajala, and Idan Habler).
• Perplexity AI Fellow (2025)

Selected Publications & Research

• IEEE HPEC 2026: STSS: Skill Trust and Signing Service for Secure AI Agent Skill Ecosystems
• IEEE HPEC 2026: Proactive Security for Large-Scale AI Inference: Adaptive Container Orchestration in Kubernetes — with Zsolt Németh (R6 Security).
• IEEE HPEC 2026: TriShield: A Unified ASIC Architecture for Composable TEE, FHE, and Differential Privacy with Leakage-Aware Formal Guarantees
• IETF Internet-Draft (ANS): Agent Name Service (ANS): A Universal Directory for Secure AI Agent Discovery and Interoperability — co-authored draft proposing a DNS-inspired, PKI-based registry for secure AI agent discovery and identity verification across protocols such as A2A, MCP, and ACP.
• IETF Internet-Draft (ANS v2): Agent Name Service v2 (ANS): A Domain-Anchored Trust Layer for Autonomous AI Agent Identity — successor draft anchoring agent identity to verified DNS domains, with tiered Bronze/Silver/Gold trust levels and an IETF SCITT-aligned transparency log.
• arXiv — ANS: Agent Name Service (ANS): A Universal Directory for Secure AI Agent Discovery and Interoperability — the research paper behind the ANS IETF draft; supported and endorsed by the OWASP GenAI Security Project. Best Paper Award, IEEE ICAIC-2026.
• Industry Adoption of ANS: Industry players including GoDaddy, Cloudflare, and Hedera (Hashgraph) are engaging with the Agent Name Service (ANS), the IETF Internet-Draft proposal co-authored by Akram Sheriff — see The New Stack, Cloudflare & GoDaddy partnership (LinkedIn), Hedera community discussion (Reddit), and GoDaddy collaboration (LinkedIn).
• arXiv — ACNBP: Agent Capability Negotiation and Binding Protocol (ACNBP) — a protocol built on ANS for secure capability discovery, negotiation, and binding between heterogeneous AI agents, evaluated with the MAESTRO threat-modeling framework.
• arXiv — RF Fingerprinting: Hamiltonian-Inspired Attention Mechanism for Scalable RF Transmitter Fingerprinting — co-authored with Chitraksh Singh and Monisha Dhanraj (Frondeur Labs); introduces a physics-informed attention architecture with norm-preserving, Störmer–Verlet leapfrog value dynamics, achieving 99.12% same-day and 61.64% large-scale (150-transmitter) accuracy on the WiSig RF fingerprinting benchmark.
• OWASP AIVSS: AI Vulnerability Scoring System — Founding Member; a structured, quantifiable methodology to identify, assess, and mitigate vulnerabilities specific to Agentic AI and other AI systems.
• OWASP Agentic Skills Top 10 (AST10): Agentic Skills Top 10 — Major Contributor; the first comprehensive security framework documenting the 10 most critical risks in AI agent skills across major agent platforms.
• TechRxiv Preprint: FATA: A Framework-Agnostic, Task-Agnostic Agentic AI Platform for Serverless Multi-Agent Orchestration — proposes a framework-agnostic, task-agnostic platform for orchestrating multi-agent AI systems on serverless infrastructure.
• Cisco Security Whitepaper: Securing Vector Databases — co-authored with Omar Santos; covers embeddings, RAG architecture, and encryption techniques (searchable, homomorphic, secure multiparty computation) for vector database security.
• Cisco Security Whitepaper: Securing AI/ML Ops: Strategies for Protecting AI Training and Finetuning Environments — co-authored with Omar Santos, Mustafa Kadioglu, Sushma Mahadevaswamy, and Dr. Gaowen Liu; covers AI/ML Ops components, training and finetuning environment threat modeling, NIST SSDF applied to AI, identity and access management (RBAC/ABAC/Zero Trust), and supply chain security.
• Patent Disclosure: Dynamic LLM Agent Metadata Manifest-Based Discovery of Agents in an LLM Agentic Application Platform — Google Technical Disclosure Commons publication on manifest-based agent discovery for framework-agnostic agentic platforms.
• Patent Disclosure: Techniques for Deriving an LLM Agent Trust Score for Dynamically Triggering Human-in-the-Loop (HIL) Feedback in Realtime for an LLM Agentic Workflow (published January 7, 2025) — Google Technical Disclosure Commons publication on an adaptive agentic trust score for multi-agent systems that automatically triggers human-in-the-loop oversight in real time.
• U.S. Patent No. 11,523,314: Triggering Client Roaming in a High Co-Channel Interference Environment (Cisco Technology, Inc.; issued December 2022; with Xiaoguang Jason Chen, Jun Liu, Robert Edgar Barton, and Jerome Henry) — a first-of-its-kind approach in the wireless networking industry, using ML (polynomial/multi-variable regression) on live RSSI and radio telemetry to predict the optimum client-roaming trigger threshold in real time, replacing a static, manually-tuned threshold that caused dropped packets, poor roaming behavior, and costly manual RF site surveys at scale. Productizing this AI/ML-driven roaming feature resolved an escalating, NPS-impacting customer issue, eliminated the need for a dedicated field engineer (~$70K/year OPEX savings per customer), and became a reference differentiator that helped win new and competitive deals, growing product revenue over subsequent quarters.
• U.S. Patent No. 12,211,625: Systems and Methods for Detecting and Tracking Infectious Diseases Using Sensor Data (Cisco Technology, Inc.; issued January 2025; with Hazim Hashim Dahir and Thomas Szigeti) — network and sensor-based system for detecting, monitoring, and tracking the spread of infectious diseases such as COVID-19 using environmental and localized sensor data (infrared, RTLS, WiFi AP connectivity, HVAC integration). Researched and built at the height of the COVID-19 pandemic out of a personal passion to contribute to society with technology, and carried into a product within Cisco's DNA Spaces portfolio.
• Voice AI Infrastructure Patents (Filed): Strategic, industry-thought-leadership patent filings in voice agentic AI infrastructure — Voice AI Inference System (Appl. No. 19/279,667, filed July 2025); Voice Agentic Infra and Inference System (Appl. No. 19/279,653, filed July 2025); and vLLM-Based Voice Agentic AI Inference — Dynamic Traffic Management and Opex Cost Optimization for Voice Agentic Tool Calls at Runtime (Appl. No. 961,858, filed January 2026).
• Cisco Blog: Emerging Trends in IoT Gateway and Edge Application Management in a Cloud Native Paradigm — outlines six critical pillars of cloud-native IoT management architecture (scalability, high-frequency and low-latency data processing, robust pipelines, protocol variety, and cloud-native messaging) and Cisco's IoT Operations Dashboard.
• Cisco Outshift Blog: Harnessing the Power of Intent-Driven Internet of LLM Agents (IIOAs) for Enterprise IT — perspective on intent-driven LLM agent orchestration for enterprise IT environments.
• Referenced In: A New Identity Framework for AI Agents — Omar Santos (Cisco) credits Akram Sheriff as a co-architect of ANS in this proposal for a zero-trust identity framework (DIDs, Verifiable Credentials, ZKPs) for autonomous agents.
• Medium: Securing and Scaling Agentic AI Inference Workloads in a Multi-Cloud Environment
• Medium: Navigating Agent Drift: A Technical Guide to Sustaining AI Agent Performance
• Medium: What is an Agentic Graph-Based AI System and Its Architectural Design Characteristics
• Open Source: MCP SIP Gateway — SIP/WebRTC telephony gateway with OpenAI Realtime API integration and MCP tool-calling support.
• Open Source: FLAME (Cisco Systems) — a federated learning platform for edge deployments, with a Go-based control-plane service and a Python data-plane library supporting FedAvg, FedYogi, FedAdam, FedProx, hierarchical, and hybrid FL algorithms.
• Springer Book Chapter (2010): Comparison of Cascaded LMS-RLS, LMS and RLS Adaptive Filters in Non-Stationary Environments — in Novel Algorithms and Techniques in Telecommunications and Networking (Springer); co-authored with Bharath Sridhar, Dr. K.A. Narayanan Kutty, and S. Sathish Kumar, proposing a cascaded LMS-RLS prediction filter for faster convergence in non-stationary signal environments.
• B.Tech UG Micro-Projects (2006–08): A series of real-world embedded projects spanning Microchip MPLAB IDE-based MCU programming and ATMEL microcontrollers, plus a Xilinx FPGA-based implementation of an edge computer-vision image-detection pipeline.
• B.Tech Undergrad Project (2008): DSP & Image Signal Processing — Eigenvalue-based multi-filter approach for face recognition, implemented with the MATLAB Neural Network Toolbox.
• B.Tech Final Year Project (2009): Blind channel estimation and semi-blind channel estimation techniques in MIMO OFDM wireless systems, using heuristics-based and ML-based (MATLAB Neural Network Toolbox) approaches in non-stationary environments.

Selected Links

• Google Scholar
• Google Patents — Akram Sheriff
• Cisco Blog
• GitHub
• Semantic Scholar
• Medium
• ORCID
• DBLP
• ResearchGate
• Tracxn — Co-Founder/CTO, Stealth-Mode Startup
• MIT NANDA Team
• NANDA Agentic AI Research Webinar Presentation — Akram Sheriff, NANDA Researcher
• StreamYard Talk
• Embedded Vision Summit — Speaker
• DevNetwork Advisory Board
• YouTube
• Cisco: Federated ML Architecture for Computer Vision in the IoT Edge (Preview)
• OASIS + Cisco AI Security Summit 2023 — LLM Security Paradigms (RTP, NC)
• UC Berkeley RDI Agentic AI Summit 2025 — Project Demo: Agentic AI Security (IAM)

Collaborators

I have been fortunate to work with some really smart folks at CMU University:

• Sarah Santos — ssantos@andrew.cmu.edu
• Jonathan Aldrich — jonathan.aldrich@cs.cmu.edu

Testimonials

"I am writing to offer my highest recommendation for Akram Sheriff, whose visionary leadership and extensive expertise in artificial intelligence have significantly advanced the field and inspired many within our community.

Akram has been at the forefront of AI development, successfully spearheading multiple projects that have not only showcased innovation but also delivered practical solutions to complex problems. Through his strategic foresight and profound technical acumen, Akram has led his team to achieve remarkable milestones, often exceeding project goals and expectations.

As an influencer in AI, Akram has a proven track record of thought leadership, contributing valuable insights in leading conferences and seminars around the world. His ability to articulate the nuances of AI, as well as its ethical implications, has made him a sought-after speaker and consultant. Akram is adept at simplifying complex concepts to engage a wider audience, thereby fostering a deeper understanding and appreciation for the potential of AI across various sectors.

Akram's commitment to mentoring emerging talent in AI is commendable. Akram has invested considerable time and resources in nurturing the next generation of AI professionals, emphasizing the importance of interdisciplinary collaboration and continuous learning. This dedication to growth and education within the AI community has helped shape an environment that encourages innovation and excellence.

Akram's charismatic leadership style, combined with a collaborative approach, has cultivated a team culture that is inclusive, forward-thinking, and highly productive. Akram not only leads by example but also empowers team members to take initiative and contribute their best work.

It is without reservation that I recommend Akram Sheriff as a foundational leader in the field of AI. His ongoing contributions are not only propelling the industry forward but are also establishing new benchmarks for success and responsible AI implementation."

— Ray Jackson, former Senior Director, Cisco Systems
On Akram's leadership driving GenAI initiatives at Cisco Systems, 2022–2024